減重不只是瘦身！研究揭示：控制脂肪生成有望逆轉脂肪肝

本翻譯僅作學術交流用，無商業意圖，請勿轉載，如有疑議問請來信

研究顯示，體重減輕約10%能讓脂肪肝患者的肝內三酸甘油脂（IHTG）大幅下降超過75%，並明顯改善胰島素敏感性。這一變化並非單純來自總熱量或脂肪攝取的減少，而是與肝臟內新生脂肪生成（DNL）下降達67%密切相關，顯示DNL是脂肪肝惡化的重要驅動因子，同時也是高度可調控的治療目標。

MASLD 中的體重減輕恢復了肝臟脂肪酸來源的平衡

Weight loss in MASLD restores the balance of liver fatty acid sources

Lambert JE, Ramos-Roman MA, Valdez MJ, Browning JD, Rogers T, Parks EJ. Weight loss in MASLD restores the balance of liver fatty acid sources. J Clin Invest. 2025;135(9):e174233. Published 2025 May 1. doi:10.1172/JCI174233

https://pubmed.ncbi.nlm.nih.gov/40309768/

摘要 Abstract

背景。脂肪生成在代謝功能障礙相關的脂肪肝疾病（MASLD）中對肝內三酸甘油脂（IHTG）的病理性積累貢獻重大。由於肝臟脂肪生成對能量攝入高度敏感，我們假設體重減輕所引起的 MASLD 逆轉機制將由脂肪生成途徑的顯著減少驅動。
BACKGROUND. Lipogenesis contributes substantially to the pathological accumulation of intrahepatic triacylglycerol (IHTG) in metabolic dysfunction–associated steatotic liver disease (MASLD). Since hepatic lipogenesis is highly sensitive to energy intake, we hypothesized that mechanisms of MASLD regression induced by weight loss would be driven by a marked reduction in the lipogenic pathway.

方法。超重的成年人具有高肝脂肪（HighLF；n = 9；IHTG ≥ 5.6% 由 1 H-磁共振光譜測量）或低（正常）肝脂肪（LowLF；n = 6；IHTG < 5.6%）接受了為期 6 個月的飲食諮詢，並在住院研究中進行了全面的代謝表型分析，捕捉了空腹和進食狀態。使用多種穩定同位素來評估脂肪生成、游離脂肪酸（FFAs）和飲食脂肪對 IHTG 的貢獻。
METHODS. Overweight adults with high liver fat (HighLF; n = 9; IHTG ≥ 5.6% measured by 1H-magnetic resonance spectroscopy) or low (normal) liver fat (LowLF; n = 6; IHTG < 5.6%) received dietary counseling for 6 months and underwent comprehensive metabolic phenotyping during inpatient studies that captured fasting and fed states. Multiple stable isotopes were used to assess the contribution of lipogenesis, free fatty acids (FFAs), and dietary fat to IHTG.

結果。體重減輕（–10% ± 2%）使 MASLD 患者的 IHTG 降低（19.4% ± 3.6%降至 4.5% ± 2.1%，P < 0.001）。胰島素敏感性顯著改善（46%，P < 0.01），而來自脂肪組織的空腹 FFAs 流量沒有變化。VLDL-三酸甘油脂（VLDL-TG）濃度下降了 38%（P = 0.02），這是由於脂肪生成貢獻減少了 67%（P = 0.02），而 FFAs 和飲食脂肪對 VLDL-TG 的絕對貢獻沒有變化。脂肪生成的減少與 IHTG 的減少顯著相關。
RESULTS. Body weight loss (–10% ± 2%) reduced IHTG in individuals with MASLD (19.4% ± 3.6% to 4.5% ± 2.1%, P < 0.001). Insulin sensitivity improved significantly (46%, P < 0.01), while fasting FFA flux from adipose tissue was not different. VLDL-triacylglycerol (VLDL-TG) concentrations fell by 38% (P = 0.02) because of a 67% reduction in contribution from lipogenesis (P = 0.02), whereas the absolute contributions from FFAs and dietary fat to VLDL-TG were not different. Reduced lipogenesis was significantly associated with loss of IHTG.

結論。這些數據強調了脂肪生成在 MASLD 病理中的主要作用，並突顯了通過治療策略控制這一路徑的重要性。
CONCLUSION. These data underscore the primary role of lipogenesis in MASLD pathology and highlight the importance of controlling this pathway through treatment strategies.

引言 Introduction

代謝功能障礙相關的脂肪肝病（MASLD）的盛行率正在隨著人口中肥胖和胰島素抵抗率的上升而上升（1–4）。在接下來的 10 年內，MASLD 將成為發達國家中最常見的肝病（5–7）。用於肝內三酸甘油脂（IHTG）合成的脂肪酸來源包括來自脂肪組織脂解的血漿游離（非酯化）脂肪酸（FFAs）、膳食脂肪以及通過新生脂肪生成（DNL）在肝臟中製造的脂肪酸。使用正電子發射斷層掃描（PET）成像的示蹤劑研究估計，肝臟吸收流向該器官的脂肪酸的比例保持在一個恆定的範圍內（約 20%–25%）（8）。使用標記技術顯示，來自脂肪組織到肝臟的無限制 FFAs 流量被證明是 IHTG 的主要脂肪酸來源，因此，MASLD 中的脂肪組織胰島素抵抗可能促進了過量的 IHTG（9, 10）。然而，我們和其他研究者（11–13）已經顯示，即使與沒有 MASLD 但具有相似程度的肥胖和胰島素抵抗的個體相比，MASLD 患者的 DNL 也顯著升高（11）。 此外，我們已顯示隨著組織學評估的肝病惡化，脂肪酸合成（DNL）以漸進的方式增加（14），而抑制脂肪生成的治療可降低肝內脂肪（IHTG）（15, 16），而且這種減少可以在短短 10 天內發生（17）。

The prevalence of metabolic dysfunction–associated steatotic liver disease (MASLD) is rising in tandem with increasing rates of obesity and insulin resistance in the population (1–4). Over the next 10 years, MASLD will become the most common liver disease in developed countries (5–7). Sources of fatty acids used for intrahepatic triacylglycerol (IHTG) synthesis include plasma free (non-esterified) fatty acids (FFAs) derived from adipose lipolysis, dietary fat, and fatty acids made in the liver via de novo lipogenesis (DNL). Tracer studies using PET imaging have estimated that the liver takes up a constant proportion (~20%–25%) of the fatty acids that flow to the organ (8). Using labeling techniques, unrestrained FFA flux from adipose tissue to the liver was shown to provide the primary source of fatty acids to IHTG, and therefore adipose insulin resistance in MASLD likely contributes to excess IHTG (9, 10). However, we and others (11–13) have shown that DNL is also markedly elevated in individuals with MASLD, even when compared with individuals without MASLD but with a similar degree of obesity and insulin resistance (11). Further, we have shown that DNL is increased in a graded fashion as histologically assessed liver disease worsens (14), that treatment with inhibitors of lipogenesis lowers IHTG (15, 16), and that this reduction can occur in as little as 10 days (17).

為了研究脂肪肝中脂質積累的機制，大多數研究集中在測量三種脂肪酸來源之一（脂肪組織、飲食或 DNL）。不幸的是，這種策略無法支持發現控制肝細胞 VLDL-三酸甘油脂（VLDL-TG）組裝的過程。在 VLDL 通過內質網和高基體的成熟過程中，三酸甘油脂會被添加到顆粒中，這些三酸甘油脂可以來自三種脂質來源中的任何一種，或來自細胞液滴中儲存的三酸甘油脂（18）。通過使用多重同位素，我們已經顯示出來自不同來源的脂肪酸在肝臟中的處理方式不同（例如，一部分游離脂肪酸會立即以 VLDL 的形式重新分泌，而新生脂肪酸似乎在分泌之前會進入肝臟儲存池，等等），並且貢獻於儲存的 IHTG 的脂肪酸來源在血漿中的 VLDL-TG 的分數來源中得到了反映（12, 14）。

To investigate mechanisms of lipid accrual in fatty liver, most studies have focused on the measurement of one of the 3 fatty acid sources (adipose, diet, or DNL). Unfortunately, this strategy does not support discovery of the processes controlling hepatocellular VLDL-triacylglycerol (VLDL-TG) assembly. During maturation of VLDL through the endoplasmic reticulum and Golgi, TGs are added to the particle that can be derived from any of the 3 lipid sources, or from TGs stored in cellular droplets (18). Using multiple isotopes, we have shown that fatty acids from different sources are processed differently in the liver (e.g., a portion of FFAs are immediately re-secreted in VLDL, while de novo fatty acids appear to enter a liver storage pool before secretion, etc.) and that the fatty acid sources contributing to stored IHTG are reflected in the fractional sources found in VLDL-TG out in the plasma (12, 14).

目前，減重仍然是治療 MASLD 最安全且最有效的方法（19）。減重干預主要針對肥胖和/或 2 型糖尿病患者，並在 IHTG 方面取得了顯著的減少。例如，IHTG 水平可以從基線的 6%降低到減重 6 週後的 2%，或從基線的 12%降低到減重 7 至 16 週後的 2%（20, 21）。雖然已經確立減重可以改善肝臟健康（如 Balakrishnan 等人所述，參考文獻 22），但很少有研究量化減重過程中肝脂肪消退的代謝機制，或全面描述這一過程中脂質代謝的變化。

Currently, weight loss remains the safest and most effective treatment for MASLD (19). Weight loss interventions have focused on individuals with obesity and/or type 2 diabetes and achieved impressive reductions in IHTG. For example, IHTG levels can be reduced from 6% at baseline to 2% after 6 weeks of weight loss, or from 12% at baseline to 2% after 7–16 weeks (20, 21). Although it is well established that weight loss can improve liver health (as reviewed by Balakrishnan et al., ref. 22), few studies have quantitated the metabolic mechanisms responsible for resolution of liver fat via weight loss or comprehensively characterized changes in lipid metabolism during this process.

在過去的脂蛋白脂肪酸來源研究中，研究人員比較了高 IHTG 和低 IHTG 受試者的特徵，但低 IHTG 的比較組通常具有較低或沒有胰島素抗性，或年齡較輕且體重較輕。我們自己過去的數據來自於相對瘦的低 IHTG 受試者（23, 24）。在 Klein 及其同事的優雅研究中，6 名沒有肥胖的個體減少了 10%的體重，顯示出脂蛋白-TG 中的 DNL 減少了 35%，而 IHTG 減少了 50%（25）。然而，沒有標記飲食的情況下，尚不清楚體重減輕如何改變所有脂肪酸來源的肝內脂質處理。在這裡，我們試圖比較 IHTG 水平差異很大的組別，同時匹配外周胰島素敏感性、年齡、體重和血脂水平。DNL 是一條相對靈活的生化途徑，與 IHTG 的大小相關（25），對能量/碳水化合物限制高度反應，並在胰島素敏感性改善時減少（26–29）。因此，我們假設一個旨在解決脂肪肝的減重計劃將主要通過減少肝臟 DNL 來實現。 我們的目標是了解 (a) 這些減少將如何影響肝臟三酸甘油脂的其他來源，以及 (b) 對於低和高肝臟三酸甘油脂的受試者，這些機制是否會相似。在兩組具有相當基線胰島素敏感性的情況下，干預的結果將更直接反映能量限制的孤立效果，而不受不同程度的外周胰島素抗性影響結果。我們發現，體重減輕引起的肝臟脂質利用變化導致了相似的去新生脂肪酸合成降低和更大的肝臟游離脂肪酸用於三酸甘油脂合成，且在高肝臟三酸甘油脂的受試者中觀察到更大的變化幅度。非去新生脂肪酸來源的絕對使用量與肝臟三酸甘油脂的降低無關。高三酸甘油脂血症主要通過降低去新生脂肪酸合成和改善脂蛋白清除來改善。考慮到能量限制對減少脂肪酸合成的強大影響，結果支持減少去新生脂肪酸合成在促進血漿游離脂肪酸用於極低密度脂蛋白三酸甘油脂合成和分泌中的主要作用，進一步導致肝臟脂肪的減少。 這些結果強調了 DNL 在 MASLD 發展中的病理性，但又是高度可調整的角色。

In past studies of lipoprotein fatty acid sources, investigators have compared characteristics of subjects with high IHTG and low IHTG, but the low-IHTG comparison groups had lower or no insulin resistance or were younger and leaner. Our own past data from those with low IHTG were derived from relatively lean subjects (23, 24). In the elegant study of Klein and colleagues, 6 individuals without obesity who lost 10% body weight demonstrated a 35% reduction in DNL present in lipoprotein-TG, and a 50% reduction in IHTG (25). However, without labeling of the diet, it is unknown how intrahepatic lipid processing of all fatty acid sources was altered by weight loss. Here, we sought to compare groups with widely varying IHTG levels while matching for the level of peripheral insulin sensitivity, age, body weight, and blood lipids. DNL is a relatively flexible biochemical pathway, correlates with the magnitude of IHTG (25), is highly responsive to energy/carbohydrate restriction, and is reduced with improvement of insulin sensitivity (26–29). Therefore, we hypothesized that a weight loss program designed to resolve fatty liver would do so primarily by reducing hepatic DNL. Our goal was to understand (a) how these reductions would impact the other sources of liver-TG, and (b) whether the mechanisms would be similar for subjects with low and high IHTG. With equivalent baseline levels of insulin sensitivity in the 2 groups, the results of the intervention would more directly reflect the isolated effect of energy restriction without the confounding effect of different levels of peripheral insulin resistance influencing the results. We found that weight loss–induced alterations in hepatic lipid utilization resulted in similar lowering of DNL and greater hepatic FFA use for TG synthesis, with a greater magnitude of changes observed in the high-IHTG subjects. The absolute use of non-DNL sources was not related to lowering of IHTG. Hypertriacylglycerolemia was improved primarily through reductions in DNL and improved lipoprotein clearance. Given the strong effect of energy restriction to reduce fatty acid synthesis, the results support a principal role of reduced DNL to lead to greater utilization of plasma FFAs for VLDL-TG synthesis and secretion, further leading to reductions in liver fat. These results underscore the pathological, yet highly modifiable, role of DNL in MASLD development.

結果 Results

在之前的出版物中（11），我們分析了 24 名非糖尿病的代謝症候群患者的代謝變數，他們有升高的肝酵素和懷疑的 MASLD，並且擁有低肝脂肪（<5.6%；低肝脂肪）或高肝脂肪（>5.6%；高肝脂肪）（請參見補充圖 1 中的招募流程；補充材料可在線獲得，網址：https://doi.org/10.1172/JCI174233DS1）。在那次分析中，受試者在胰島素抵抗方面進行了匹配，而高肝脂肪的受試者的一個顯著特徵是肝臟脂肪生成（DNL）是低肝脂肪的 2 倍（11）。在這些基線研究之後，16 名受試者同意參加為期 6 個月的減重干預；其中 15 名完成了干預並被納入分析（1 名因遵從性差而被排除）。本研究的目的是通過減少總能量攝入來實現減重，這是通過改善飲食質量來實現的，具體方法是減少簡單糖的攝入並增加全食物的消費。 透過實現有意義的體重減輕，我們的目標是測試能量限制對脂肪酸流動和 VLDL-TG 脂肪酸來源的影響（主要終點），識別與 IHTG 和 VLDL-TG 減少相關的代謝機制（次要終點），並比較起始時肝臟脂肪高或低的個體的體重減輕效果（圖 1B；次要終點）。入院 1 是頻繁取樣的靜脈葡萄糖耐受測試，入院 2 旨在全面評估脂肪酸代謝（圖 1B）。根據設計，這些組別在 IHTG 的基線含量上有所不同（低肝脂肪 2.1% ± 0.9%，n = 6，對比高肝脂肪 19.4% ± 1.2%，n = 9，P < 0.001），但在靜脈葡萄糖耐受測試的結果、年齡（45 ± 6 歲和 50 ± 9 歲，P = 0.292）、體重、飲食攝入、人體測量和空腹生化指標上是匹配的（表 1）。然而，胰島素濃度（P = 0.058）和胰島素抵抗的靜態標記（胰島素抵抗的穩態模型評估[HOMA-IR]和脂肪組織胰島素抵抗[AdipoIR]，一種脂肪組織胰島素敏感性的指標）在高肝脂肪組中仍然傾向於較高（表 2）。

In a previous publication (11), we analyzed metabolic variables in 24 non-diabetic subjects with metabolic syndrome, elevated liver enzymes, and suspected MASLD, who possessed either low liver fat (<5.6%; LowLF) or high liver fat (>5.6%; HighLF) (see recruitment flow in Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI174233DS1). In that analysis, subjects were matched for insulin resistance, and a distinctive characteristic of those with HighLF was a 2-fold greater hepatic DNL (11). After those baseline studies, 16 of the subjects agreed to enter a 6-month weight loss intervention; of those, 15 completed the intervention and were included for analysis (1 subject was excluded because of poor compliance). The aim of the present study (Figure 1A) was to produce weight loss by reducing overall energy intake, achieved through improvement of diet quality by reduction of intake of simple sugars and increasing consumption of whole foods. By bringing about meaningful weight loss, our goal was to test the influence of energy restriction on fatty acid flux and sources of VLDL-TG fatty acids (primary endpoints), to identify metabolic mechanisms associated with reductions in both IHTG and VLDL-TG (secondary endpoints), and to compare weight loss effects in individuals who started with either high or low liver fat (Figure 1B; secondary endpoints). Admission 1 was a frequently sampled i.v. glucose tolerance test, and admission 2 was designed to comprehensively assess fatty acid metabolism (Figure 1B). By design, the groups differed in their baseline content of IHTG (LowLF 2.1% ± 0.9%, n = 6, vs. HighLF 19.4% ± 1.2%, n = 9, P < 0.001), but were matched for the results of the i.v. glucose tolerance tests, age (45 ± 6 years and 50 ± 9 years, P = 0.292), body weight, dietary intakes, anthropometrics, and fasting biochemistries (Table 1). However, insulin concentrations (P = 0.058) and static markers of insulin resistance (homeostatic model assessment for insulin resistance [HOMA-IR] and adipose insulin resistance [AdipoIR], an index of adipose insulin sensitivity) still tended to be higher in the HighLF (Table 2).

圖 1 Figure 1

六個月的研究設計和住院協議，以量化脂肪酸和三酸甘油脂的代謝。(A) 招募受試者參加為期六個月的飲食研究，以確定體重減輕引起的肝臟脂肪改善的代謝機制，包括胰島素代謝（在入院時通過頻繁取樣的胰島素修飾靜脈注射葡萄糖耐受測試進行評估）和脂肪酸代謝。(B) 在體重減輕前後，脂肪酸代謝是在一個過夜同位素輸注研究中測量的。在住院研究前的 10 天內進行了重水的口服給藥，並將脂肪酸同位素納入晚餐中，並靜脈注射以追蹤脂肪組織游離脂肪酸的流量及 VLDL-TG 棕櫚酸的來源。對 24 小時研究的數據進行分析，以評估餐後脂質代謝、夜間脂質流量，以及當受試者禁食額外 4 小時後，這些過程如何改變，隨後進餐。CTRC，臨床與轉化研究中心；FFA，游離脂肪酸。

Six-month study design and inpatient protocol to quantitate fatty acid and TG metabolism. (A) Subjects were recruited to participate in a 6-month dietary study to determine the metabolic mechanisms of weight loss–induced improvements in liver fat, including insulin metabolism (assessed in admission 1 via a frequently sampled, insulin-modified i.v. glucose tolerance test) and fatty acid metabolism. (B) Before and after weight loss, fatty acid metabolism was measured during an overnight isotope infusion study. Oral dosing with deuterated water occurred for 10 days before the inpatient study, and fatty acid isotopes were incorporated into the evening meal and infused i.v. to track adipose FFA flux and the sources of VLDL-TG palmitate. Data from the 24-hour study were analyzed to assess post-meal lipid metabolism, nighttime flux of lipids, and how these processes changed when the subjects fasted an additional 4 hours, followed by a meal. CTRC, Clinical and Translational Research Center; FFA, free fatty acids.

表 1 低脂肪和高脂肪受試者在減重前後的特徵 表 2 低肝脂肪和高肝脂肪在減重前後的胰島素敏感性變化

體重減輕對肝臟脂肪的影響。表 1 顯示了受試者在基線和 6 個月干預結束時的飲食組成。整體能量攝入的減少和飲食質量的改善導致所有宏量營養素（脂肪、蛋白質、碳水化合物和糖）的攝入量減少，而纖維攝入量保持不變。整體而言，受試者飲食的組成特徵是來自脂肪的能量百分比減少（基線的 37% ± 2%降至隨訪的 31% ± 2%）、總糖（21% ± 2%降至 17% ± 1%）和添加糖（14% ± 2%降至 6% ± 1%），而來自蛋白質的能量則增加（17% ± 1%升至 23% ± 1%）。透過圖 2A 中相應的體重逐步減輕，可以觀察到逐步和系統性減少食物攝入的效果。低脂肪組（LowLF）和高脂肪組（HighLF）均出現約 10%的顯著體重減輕（圖 2A 和表 1）。 體重減輕的主要效果（P < 0.001）不受肝脂肪組別交互作用的影響（P = 0.229）。體重減輕導致高脂肪組（HighLF）內部肝臟脂肪（IHTG）顯著減少 75.6% ± 4.0%（P < 0.001，顯著的交互作用效果為體重減輕 × 肝脂肪組別；圖 2B），並使大多數受試者的 IHTG 正常化（<5.6%）。高脂肪組的 IHTG 降低至與低脂肪組（LowLF）相似的水平，使我們能夠確定這兩組在隨訪時肝臟脂肪酸的流量是否相似。兩組的肝酶在體重減輕後均顯著下降（表 1）。在 9 名高脂肪組受試者中，有 4 名受試者在基線時接受了診斷性（醫療指示）肝活檢，並在體重減輕後進行了重複活檢。這些受試者的結果包含在補充圖 2 中，以顯示體重減輕對肝損傷的影響。 例如，基線時輕度肝損傷的受試者 4（補充圖 2A）經歷了 11%的體重減輕，這與胰島素敏感性的加倍以及丙氨酸轉氨酶（減少 83%）、血漿三酸甘油脂（減少 50%）和肝臟脂肪（從 27.3%降至 1.8%，通過磁共振光譜法[MRS]評估）減少有關，並伴隨有組織學證據顯示脂肪變性和纖維化的消退。

Effect of weight loss on liver fat. Table 1 shows the dietary composition of the subjects at baseline and at the end of the 6-month intervention. Reductions in overall energy intake and improvement in dietary quality brought about reductions in the intake of all macronutrients (fat, protein, carbohydrate, and sugars), while fiber intake was maintained. Overall, the composition of the subjects’ diets was characterized by a reduction in percentage energy from fat (37% ± 2% at baseline to 31% ± 2% at follow-up), total sugars (21% ± 2% to 17% ± 1%), and added sugars (14% ± 2% to 6% ± 1%), while the energy from protein increased (17% ± 1% to 23% ± 1%). The effects of the stepwise and systematic reduction in food intake can be observed through the corresponding progressive loss of weight in Figure 2A. Similar significant reductions in body weight of about 10% occurred in both LowLF and HighLF groups (Figure 2A and Table 1). The main effect of weight loss (P < 0.001) was not affected by a liver fat group interaction (P = 0.229). Weight loss resulted in a significant 75.6% ± 4.0% relative reduction in IHTG in the HighLF group (P < 0.001 for a significant interaction effect of weight loss × liver fat group; Figure 2B) and resulted in normalization of IHTG (<5.6%) in the majority of subjects. The lowering of IHTG in the HighLF group to levels similar to those in the LowLF group allowed us to determine whether the 2 groups had similar fluxes of liver fatty acids at follow-up. Liver enzymes fell significantly with weight loss in both groups (Table 1). Of the 9 HighLF subjects, 4 subjects underwent diagnostic (medically indicated) liver biopsies at baseline and also had repeat biopsies after weight loss. Results from these subjects are included in Supplemental Figure 2 to show the effect of weight loss on liver injury. For example, subject 4 with mild liver injury at baseline (Supplemental Figure 2A) experienced an 11% weight loss, which was associated with a doubling of insulin sensitivity and reductions in alanine transaminase (–83%), plasma TGs (–50%), and liver fat (27.3% to 1.8%, assessed by magnetic resonance spectroscopy [MRS]) along with histological evidence of resolution of steatosis and fibrosis.

圖 2 Figure 2

改善禁食和餐後全身代謝。在入院第 2 天（圖 1B），受試者在第 1 天的 1800 小時進食標準化晚餐，隨後禁食至第 2 天的中午（1200 小時）。這一方案有兩個原因：它使我們能夠測試受試者適應急性延長禁食的能力，並且為 VLDL-TG 的穩定同位素標記提供了足夠的時間以達到穩態。在第 2 天的 1200 小時，提供與晚餐相同（成分和能量相同）的午餐，並在 1800 小時之前採集額外的血樣，以評估在延長禁食後進食的代謝反應。將描述 24 小時測試期間的四個階段，並通過在四個研究時間段（餐後、夜間、延長禁食和午餐）之間的轉變來觀察代謝的變化。

Improved fasting and post-meal whole-body metabolism. During admission 2 (Figure 1B), subjects consumed a standardized dinner at 1800 hours on day 1, followed by fasting until noon (1200 hours) on day 2. This protocol was used for 2 reasons: it allowed us to test the ability of the subjects to adapt to an acute extended fast and allowed enough time for stable isotope labeling of VLDL-TG to come to a steady state. At 1200 on day 2, a lunch meal identical to the evening dinner (same composition and energy) was fed, and additional blood samples were taken until 1800 to assess metabolic responses to eating after an extended fast. Four phases of the 24-hour test duration will be described in terms of changes in metabolism observed through transitioning between fasting and fed states across the 4 study time periods (post-dinner, night, extended fast, and lunch).

血漿葡萄糖、胰島素、游離脂肪酸（FFAs）和三酸甘油脂（TGs）的晝夜節律反應如圖 3 所示，而禁食水平則在補充圖 3 中總結。晝夜節律反應通過三因素重複測量方差分析（RM-ANOVA）來分析代謝物濃度的變化，以評估（a）禁食和進食狀態之間代謝物的模式和波動，（b）體重減輕的影響，以及（c）不同肝脂肪組別觀察到的影響之間的差異。葡萄糖（圖 3，A 和 B）在晚餐後增加，隨後逐漸下降，在夜間和延長的禁食期間保持平穩，然後在午餐後再次上升，隨後再次逐漸下降（時間過程的主要效應 P < 0.001）。禁食後的葡萄糖濃度在體重減輕後並無顯著差異（補充圖 3A）。體重減輕在 24 小時內有降低血漿葡萄糖濃度的趨勢（體重減輕的效應 P = 0.063）。

Circadian responses of plasma glucose, insulin, FFAs, and TGs are shown in Figure 3, while fasting levels are summarized in Supplemental Figure 3. Circadian responses were analyzed by 3-way repeated-measures ANOVA (RM-ANOVA) for changes in metabolite concentrations in order to assess (a) the pattern and fluctuation of metabolites between fasting and fed states, (b) the effects of weight loss, and (c) any differences in observed effects by liver fat group. Glucose (Figure 3, A and B) increased following the dinner meal and progressively fell, plateaued during the night and the extended fasting period, and then rose following the lunch meal before progressively falling again (P < 0.001 for main effect of time course). Fasting glucose concentrations were not significantly different following weight loss (Supplemental Figure 3A). There was a tendency for weight loss to reduce plasma glucose concentrations over the 24-hour period (P = 0.063 for effect of weight loss).

圖 3 Figure 3

餐後胰島素濃度（圖 3，C 和 D）在晚餐後上升，隨後逐漸下降，並在過夜及延長的禁食期間保持平穩，然後在午餐後再次上升，隨後又回落至禁食水平。體重減輕顯著降低了兩組的血漿胰島素濃度，特別是在餐後的進食狀態（晚餐和午餐期間）（時間過程×體重減輕交互作用 P < 0.001；圖 3，C 和 D）。體重減輕後，餐後胰島素峰值高度較低，且更快回到基線，顯示體重減輕的主要影響是對餐後狀態的影響。這一發現與在兩組中觀察到的通過靜脈葡萄糖耐受測試（入院 1）評估的敏感性改善一致（表 2）。在兩組中，體重減輕後的禁食胰島素濃度顯著降低（P = 0.002；補充圖 3B）。在 HighLF 組中觀察到胰島素的更大減少（禁食胰島素的體重減輕×肝脂肪組交互作用效應 P = 0.075，以及 HighLF 基線到隨訪的變化 P < 0.001）。 高 LF 游離脂肪酸的減少，加上空腹胰島素減少 50%，導致 AdipoIR 顯著降低（表 2）。兩組均顯示 HOMA-IR（反映全身胰島素抗性）和處置指數（反映相對於作用的胰島素分泌）有所改善，雖然兩組的胰島素敏感性指數均有所上升，但在 LowLF 組中上升的趨勢更明顯（表 2）。

Insulin concentrations (Figure 3, C and D) rose following the dinner meal and progressively fell, plateauing overnight and for the extended fasting period, and then rose again following the lunch meal before falling back to fasting levels. Weight loss significantly reduced plasma insulin concentrations in both groups, particularly in the fed states following the meals (post-dinner and lunch periods) (P < 0.001 for time course × weight loss interaction; Figure 3, C and D). After weight loss, lower insulin peak heights following the meals and a faster return to baseline demonstrate that the effect of weight loss was to chiefly impact the postprandial state. This finding is consistent with the improved sensitivity assessed by the i.v. glucose tolerance test (admission 1) observed in both groups (Table 2). Fasting insulin concentrations were significantly lower after weight loss in both groups (P = 0.002; Supplemental Figure 3B). A greater reduction in insulin was observed in the HighLF group (P = 0.075 for weight loss × liver fat group interaction effect for fasting insulin, and P < 0.001 for change from baseline to follow-up in the HighLF). The reduction in HighLF FFAs, along with a 50% reduction in fasting insulin, resulted in a significantly greater reduction in AdipoIR (Table 2). Both groups exhibited improvement in HOMA-IR (reflecting whole-body insulin resistance) and the disposition index (reflecting insulin secretion relative to action), and while the insulin sensitivity index rose in both groups, it tended to rise more in the LowLF group (Table 2).

夜間游離脂肪酸（FFAs）的延長升高與胰島素抵抗有關（30–32）。餐後，血漿 FFAs 濃度（圖 3，E 和 F）因胰島素抑制脂肪組織脂解而下降，隨後逐漸上升，在過夜和延長禁食期間達到平穩，然後在午餐後再次下降。這些晝夜節律模式在體重減輕的情況下觀察到沒有顯著影響（圖 3，E 和 F），儘管與高脂肪組（HighLF）相比，低脂肪組（LowLF）在體重減輕後夜間 FFAs 濃度傾向於增加（P = 0.071，體重減輕×肝臟脂肪組互動）。高脂肪組的禁食 FFAs 濃度傾向於下降（P = 0.037；補充圖 3C），但低脂肪組則沒有變化。
Extended elevations in nighttime FFAs have been implicated in insulin resistance (30–32). Plasma FFA concentrations (Figure 3, E and F) fell after the meal as a result of insulin suppression of adipose lipolysis and then progressively rose, plateauing overnight and through the extended fast, before falling again in response to lunch. These circadian patterns were observed with no significant effect of weight loss (Figure 3, E and F), though overnight FFA concentrations tended to increase in the LowLF group after weight loss compared with HighLF (P = 0.071 for weight loss × liver fat group interaction). Fasting FFA concentrations tended to fall in the HighLF group (P = 0.037; Supplemental Figure 3C) but did not change in the LowLF group.

最後，血漿三酸甘油脂濃度在餐後上升，這是由於餐脂肪中攜帶三酸甘油脂的乳糜微粒進入血液，然後隨著它們在外周組織被脂解作用清除而逐漸下降。殘餘物被肝臟吸收。在這裡，兩組在減重後的整體血漿三酸甘油脂水平顯著降低（圖 3，G 和 H），導致隨訪時空腹血漿三酸甘油脂濃度降低（P = 0.006，減重的主要效應；補充圖 3D），特別是在高脂肪組（P = 0.004）。如圖 3，G 和 H 所示，減重後餐前的血漿三酸甘油脂濃度降低（如圖表中的-6 小時時間點所示），而餐後的血漿脂血症在減重後顯著降低，儘管低脂肪組顯示出更快回到空腹水平（P = 0.056，時間過程×肝脂肪組互動）。

Finally, plasma-TG concentrations rise postprandially as a result of influx of chylomicrons carrying TGs from meal fat before progressively falling as they are cleared by lipolysis at peripheral tissues. Remnants are taken up by the liver. Here, overall plasma-TG levels were significantly lower in both groups after weight loss (Figure 3, G and H), resulting in reduced fasting plasma TG concentrations at follow-up (P = 0.006 for main effect of weight loss; Supplemental Figure 3D) and particularly in the HighLF group (P = 0.004). As shown in Figure 3, G and H, pre-dinner concentrations of plasma-TG were reduced with weight loss (as seen at the –6 hour time point on the graph), and plasma lipemia after both dinner and lunch was significantly lower after weight loss, though the LowLF group demonstrated a faster return to fasting levels (P = 0.056 for time course × liver fat group interaction).

呼吸氣體分析用於評估禁食到進食狀態下，體重減輕後葡萄糖和脂肪氧化的變化是否不同（圖 4）。與高脂肪組相比，低脂肪組在禁食和進食狀態下的葡萄糖氧化均較低，體重減輕後（肝臟脂肪組互動的 P 值 = 0.079；圖 4A）。在低脂肪組的基線中，午餐攝取使葡萄糖氧化增加了 73%（從 0.85 增加到 1.47 mg/kg/min，P = 0.058），而在體重減輕後，餐後的變化相似（63%）。在高脂肪組中，基線時餐後葡萄糖氧化上升了 70%（P = 0.013），而在體重減輕後這一上升為 56%；然而，由於受試者之間的變異性較小，刺激效果更顯著（P = 0.009）。在基線時，高脂肪組的進食狀態下脂肪氧化高於低脂肪組，並且在體重減輕後傾向於保持較高（肝臟脂肪組的主要效應 P = 0.062）。

Respiratory gas analysis was used to assess whether fasting to fed state changes in glucose and fat oxidation were different following weight loss (Figure 4). Compared with the HighLF group, glucose oxidation was lower in the LowLF group, in both the fasting and fed states, after weight loss (P = 0.079 for weight loss by liver fat group interaction; Figure 4A). In LowLF subjects at baseline, the lunch consumption increased glucose oxidation 73% (from 0.85 to 1.47 mg/kg/min, P = 0.058), and after weight loss, the meal-induced change was similar (63%). In the HighLF group, meal-induced glucose oxidation rose 70% at baseline (P = 0.013), and after weight loss this rise was 56%; however, stimulation was more significant as a result of less variability between subjects (P = 0.009). Fed-state fat oxidation was higher in the HighLF group than in the LowLF at baseline and tended to remain higher after weight loss (main effect of liver fat group P = 0.062).

圖 4 Figure 4

脂肪酸釋放。穩定同位素被用於量化血漿游離脂肪酸的周轉和代謝，棕櫚酸被用作代表性脂肪酸，以評估來自脂肪組織、飲食和脂肪生成的脂肪酸來源的貢獻（圖 1）。方法部分詳細說明了用於計算血漿游離脂肪酸從脂肪組織的出現速率（RaFFA）的多個組件。測量了三個來源的血漿游離脂肪酸：（a）從脂肪組織內部脂解釋放的游離脂肪酸，（b）從乳糜微粒的 LPL 介導脂解釋放的游離脂肪酸，以及（c）在血漿中 LPL 介導的 VLDL 脂解釋放的游離脂肪酸。圖 5A 和 B 中呈現的 RaFFA 數據僅代表脂肪釋放。在夜間或延長禁食期間（0000–1200 小時）未觀察到體重減輕對 RaFFA 的影響（P = 0.562，體重減輕的主要效應，且肝脂肪組別無效應）。

Adipose fatty acid release. Stable isotopes were administered for quantitation of plasma FFA turnover and metabolism, and palmitate was used as the representative fatty acid to assess contributions from adipose-, dietary-, and lipogenesis-derived fatty acid sources (Figure 1). The Methods section details the multiple components used to calculate the rate of appearance of plasma FFAs from adipose tissue (RaFFA). Three sources of plasma FFAs were measured: (a) FFAs released from intra-adipose lipolysis, (b) FFAs liberated from LPL-mediated lipolysis of chylomicrons, and (c) LPL-mediated lipolysis of VLDL in the plasma. The RaFFA data presented in Figure 5, A and B, represent the adipose release only. No effects of weight loss on RaFFA during the night or extended fasting periods (0000–1200 hours) were observed (P = 0.562 for main effect of weight loss, and no effect of liver fat group).

圖 5 Figure 5

用於三酸甘油脂合成的脂肪酸來源。我們之前已經顯示，VLDL-TG（飲食、游離脂肪酸和去新生脂肪酸合成）中的脂肪酸來源反映了肝臟中三酸甘油脂酯化的貢獻（12、14、23、24、33）。在這裡，這些來源是根據它們對空腹 VLDL-TG 棕櫚酸的比例貢獻計算的，這些圖表反映了脂肪酸的來源（圖 5，C–H）。除了圖 5 中呈現的夜間標記模式外，午餐後（1200 小時）脂肪酸來源的變化也包括在內，以引起讀者的興趣；然而，由於這不是一個穩態系統，因為餐後乳糜微粒的流入，因此午餐後未進行來源變化的統計分析。
Fatty acid sources used for TG synthesis. We have shown previously that the fatty acid sources found in VLDL-TG (diet, FFAs, and DNL) reflect the contributions to TG esterification within the liver (12, 14, 23, 24, 33). Here, the sources were calculated based on their proportional contribution to fasting VLDL-TG palmitate, and these graphs reflect the origin of the fatty acid (Figure 5, C–H). In addition to the nocturnal patterns of labeling presented in Figure 5, the changes in fatty acid sources after lunch (1200 hours) are also included for the reader’s interest; however, because this is not a system in steady state owing to influx of chylomicrons following the meal, statistical analysis of changes in sources was not performed after lunch.

在圖 5 的 C 和 D 中，兩組在 VLDL-TG 中飲食脂肪的外觀沒有差異，且體重減輕沒有影響。圖 5 的 E 和 F 顯示，體重減輕與兩組在整個夜間 VLDL-TG 棕櫚酸中來自 FFA 池的脂肪酸比例增加有關（P < 0.001）。此外，體重減輕似乎使 LowLF 組在夜間較早達到 FFA 對 VLDL-TG 的分數貢獻的平臺。最後，在基線時，兩組在夜間來自 DNL 的 VLDL-TG 棕櫚酸比例較高（在午夜時約為 25%–30%；圖 5 的 G 和 H），然後在整個夜間和延長禁食期間逐漸下降。體重減輕顯著降低了兩組在餐後狀態（午夜時間點）和整個夜間的 DNL（體重減輕的主要效應 P = 0.002；圖 5 的 G 和 H）。體重減輕後，LowLF 組的脂肪生成抑制非常強，達到 0200 小時時低於 5%的水平。
In Figure 5, C and D, no difference was observed between groups in the appearance of dietary fat in VLDL-TG and no effect of weight loss. Figure 5, E and F, demonstrates that weight loss was associated with a greater proportion of fatty acids from the FFA pool appearing in VLDL-TG palmitate in both groups all night long (P < 0.001). In addition, weight loss appeared to cause the LowLF group to reach a plateau of the fractional contribution of FFAs to VLDL-TG earlier in the night. Lastly, at baseline, the proportion of VLDL-TG palmitate arising from DNL was higher in both groups during the night (starting at ~25%–30% at midnight; Figure 5, G and H), and then it progressively fell overnight and through the extended fast. Weight loss significantly reduced DNL in both groups in the postprandial state (midnight time point) and throughout the night (P = 0.002 for main effect of weight loss; Figure 5, G and H). After weight loss, the LowLF group exhibited a very strong suppression of lipogenesis such that it reached levels below 5% by 0200 hours.

標籤的時間模式如圖 5 所示，而圖 6 則提供了更多有關這些標籤如何進入肝臟並用於 VLDL-TG 合成的細節。例如，飲食標籤可以通過受體介導的乳糜微粒殘餘物攝取進入肝臟，或者當乳糜微粒在血漿中經過 LPL 介導的脂解時，釋放的脂肪酸可以最終附著在白蛋白上（例如，貢獻於被稱為“溢出”的過程中的 FFA 池）。溢出脂肪酸有助於 FFA 流量。在圖 6 中，這兩條飲食脂肪使用的路徑已被劃分（均以綠色表示）。僅在 HighLF 受試者中，體重減輕增加了殘餘物進入肝臟的出現以及通過 VLDL-TG 合成回收這種脂質的過程，從基線的 3.2% ± 0.6%增加到隨訪的 5.5% ± 1.2%（P = 0.035）。相比之下，來自溢出的 FFA，提供了 VLDL 棕櫚酸的 1.1%至 5.8%，在任何組別中都未受到體重減輕的影響。 關於從脂肪組織衍生的游離脂肪酸（FFAs）在極低密度脂蛋白三酸甘油脂（VLDL-TG）中的比例，這些貢獻在減重後顯著增加——在低脂肪組（LowLF）中從 67.3% ± 5.4%增加到 80.9% ± 2.5%（P = 0.037），在高脂肪組（HighLF）中從 42.9% ± 5.8%增加到 69.7% ± 6.1%（P = 0.002）（圖 6）。血漿游離脂肪酸用於 VLDL-TG 的比例增加，顯示出肝臟對於用於 VLDL-TG 輸出的血漿游離脂肪酸的相對酯化增加。事實上，計算為肝臟吸收的血漿中可用游離脂肪酸（RaFFA）所佔比例的 VLDL-TG 中游離脂肪酸的納入率，在兩組減重後均顯示出增加的趨勢（低脂肪組從 13.6% ± 2.4%增加到 19.0% ± 4.0%，高脂肪組從 14.1% ± 2.3%增加到 20.0% ± 3.6%，P = 0.053 為減重的主要效應）。

The temporal patterns of labeling are shown in Figure 5, while Figure 6 provides more detail on how those labels reached the liver and were used for VLDL-TG synthesis. For instance, dietary label can enter the liver via receptor-mediated uptake of chylomicron remnants, or, when chylomicrons undergo LPL-mediated lipolysis in the plasma, the released fatty acids can end up on albumin (e.g., contribute to the FFA pool in a process referred to as “spillover”). Spillover fatty acids contribute to FFA flux. In Figure 6, these 2 routes of dietary fat usage have been delineated (both in green). In the HighLF subjects only, weight loss increased the appearance of remnant uptake into the liver and recycling of this lipid through VLDL-TG synthesis, accounting for 3.2% ± 0.6% at baseline to 5.5% ± 1.2% at follow-up (P = 0.035). By contrast, FFA from spillover, which provided between 1.1% and 5.8% of VLDL palmitate, was not affected by weight loss in either group. With regard to the proportions of FFAs derived from adipose tissue found in VLDL-TG, these contributions were significantly increased after weight loss — from 67.3% ± 5.4% to 80.9% ± 2.5% in the LowLF group (P = 0.037), and from 42.9% ± 5.8% to 69.7% ± 6.1% (P = 0.002) in the HighLF group (Figure 6). The increase in proportional use of plasma FFAs for VLDL-TG indicated a greater relative hepatic esterification of plasma FFAs that were used for VLDL-TG export. Indeed, the rate of FFAs incorporated into VLDL-TG, calculated as a proportion of the available FFAs in plasma (RaFFA) that are taken up by the liver, showed a trend toward increasing in both groups after weight loss (LowLF 13.6% ± 2.4% to 19.0% ± 4.0% and HighLF 14.1% ± 2.3% to 20.0% ± 3.6%, P = 0.053 for main effect of weight loss).

圖 6 Figure 6

由於在這個實驗中脂肪生成標記了超過 10 天，因此 LPL 介導的標記 VLDL-TG DNL 的脂解作用可能會將 DNL 脂肪酸洩漏到血漿 FFA 池中。我們確實在血漿 FFA 中檢測到少量來自 DNL 的脂肪酸（以下稱為 FFA-DNL），因此解釋了它們在 VLDL-TG 中的出現（圖 6）。在 LowLF 組中，基線和隨訪時，來自 FFA-DNL 的 VLDL-TG 棕櫚酸水平相似（分別為 3.7% ± 1.1%對 2.6% ± 0.7%，P = 0.091）。在 HighLF 組中，這些百分比在基線和隨訪之間也沒有差異（分別為 2.9% ± 1.1%對 2.5% ± 0.9%，P = 0.750）。在 VLDL-TG 中，直接用於肝內 TG 合成的大多數新生脂肪酸來自肝臟 DNL。體重減輕顯著降低了 LowLF 組（基線 12.0% ± 3.0%對隨訪 4.8% ± 1.7%，P = 0.032）和 HighLF 組（基線 18.5% ± 2.0%對隨訪 9.9% ± 2.1%，P = 0.011）的肝臟 DNL。
Since lipogenesis was labeled over 10 days in this experiment, LPL-mediated lipolysis of labeled VLDL-TG DNL could spill over DNL fatty acids into the plasma FFA pool. We did detect small amounts of DNL-derived fatty acids in the plasma FFA (herein referred to as FFA-DNL) and thus accounted for their appearance in VLDL-TG (Figure 6). In the LowLF group, at baseline and follow-up, similar levels of VLDL-TG palmitate were derived from FFA-DNL (3.7% ± 1.1% vs. 2.6% ± 0.7%, respectively, P = 0.091). In the HighLF group, these percentages were also not different between baseline and follow-up (2.9% ± 1.1% vs. 2.5% ± 0.9%, respectively, P = 0.750). Within VLDL-TG, the majority of de novo fatty acids that were directly used for intrahepatic-TG synthesis were derived from hepatic DNL. Weight loss significantly reduced hepatic DNL in both the LowLF group (baseline 12.0% ± 3.0% vs. follow-up 4.8% ± 1.7%, P = 0.032) and the HighLF group (baseline 18.5% ± 2.0% vs. follow-up 9.9% ± 2.1%, P = 0.011).

最後，VLDL-TG 的一部分是由在同位素給予期間未被標記的脂質合成的。對於 HighLF 組，未標記部分顯著減少（P = 0.003；圖 6），這與未標記部分來自 IHTG 儲存的概念一致，而該組的 IHTG 儲存也顯著減少（圖 2）。直接比較兩個肝臟脂肪組，在研究開始時，LowLF 組來自脂肪組織用於 VLDL-TG 合成的 FFA 比例顯著高於 HighLF 組（67.3% ± 5.4% vs. 42.9% ± 5.8%，P = 0.013；圖 6），並且傾向於有較低的 DNL（12.0% ± 3.0% vs. 18.5% ± 2.0%，P = 0.081）。體重減輕在兩組中產生了類似的變化，對 VLDL-TG 棕櫚酸的比例貢獻特徵為脂肪來源 FFA 的使用增加和 DNL 的減少。總體而言，HighLF 組在體重減輕中顯示出更大的變化幅度，這導致所有用於製造 VLDL-TG 的肝臟來源正常化，以至於在隨訪時兩組之間沒有差異。

Lastly, a portion of VLDL-TG is synthesized from lipid that does not become labeled during the duration of isotope administration. For the HighLF group, the unlabeled portion was significantly reduced (P = 0.003; Figure 6), which is consistent with the concept that the unlabeled portion emanates from IHTG stores, which were also significantly reduced in this group (Figure 2). Comparing the 2 liver fat groups directly, at the beginning of the study, the LowLF group had a significantly greater proportion of FFAs from adipose used for VLDL-TG synthesis compared with the HighLF group (67.3% ± 5.4% vs. 42.9% ± 5.8%, P = 0.013; Figure 6) and tended to have lower DNL (12.0% ± 3.0% vs. 18.5% ± 2.0%, P = 0.081). Weight loss produced similar changes in the proportional contributions to VLDL-TG palmitate in both groups, characterized by an increase in the use of adipose-derived FFAs and a reduction in DNL. Overall, the HighLF group exhibited a greater magnitude of changes with weight loss, which resulted in a normalization of all hepatic sources used to make VLDL-TG such that at follow-up there was no difference between the groups.

VLDL-TG 棕櫚酸中脂質的絕對貢獻反映了肝臟 DNL 的減少和 FFA 再酯化的增加。當 6 個來源的比例值乘以總 VLDL-TG 棕櫚酸濃度（12, 24）時，VLDL-TG 棕櫚酸中空腹來源的絕對濃度可用於評估 MASLD 中存在的高脂血症原因（補充圖 4）。與分數來源類似，VLDL-TG 棕櫚酸的絕對來源在 HighLF 受試者中被標準化，使其在減重後與 LowLF 組的數量無異。IHTG 的絕對變化與高三酸甘油脂血症的顯著減少相關（r = 0.531, P = 0.042；補充圖 5A），未標記脂肪酸的數量減少（即在 VLDL-TG 中識別出更多的棕櫚酸）（r = 0.762, P = 0.010；補充圖 5B），以及來自 DNL 的較低 TG-棕櫚酸（r = 0.650, P = 0.009；補充圖 5C）。VLDL-TG 中的其他脂肪酸來源與 IHTG 減少無關。

Absolute contributions of lipids in VLDL-TG palmitate reflect reduced hepatic DNL and increased FFA re-esterification. When the proportional values of the 6 sources are multiplied by the total VLDL-TG palmitate concentration (12, 24), the absolute concentration of fasting sources in VLDL-TG palmitate can be used to assess the causes of hyperlipidemia present in MASLD (Supplemental Figure 4). Similarly to the fractional sources, the absolute sources of VLDL-TG palmitate were normalized in the HighLF subjects to make them not different from the amounts in the LowLF group after weight loss. The absolute change in IHTG was associated with a significant reduction in hypertriacylglycerolemia (r = 0.531, P = 0.042; Supplemental Figure 5A), a reduction in the amount of unlabeled fatty acids (i.e., identification of greater amounts of palmitate in VLDL-TG) (r = 0.762, P = 0.010; Supplemental Figure 5B), and lower TG-palmitate from DNL (r = 0.650, P = 0.009; Supplemental Figure 5C). None of the other fatty acid sources in VLDL-TG were associated with IHTG reduction.

周邊三酸甘油脂代謝。最後，體重減輕降低了空腹 VLDL-TG 濃度（P = 0.040；圖 7A），特別是在 HighLF 組（P = 0.008）。VLDL-TG 濃度的降低並不是由於 VLDL-TG 產生率的降低（圖 7B），而主要是由於 VLDL-TG 的部分分解率增加（P = 0.038；圖 7C），清除率也有適度增加（體重減輕的 P = 0.126；圖 7D）。在所有受試者中（圖 8A），雖然肝臟脂肪均有減少（在某些受試者中減少多達 12 倍），但 VLDL-TG 濃度的減少幅度較小（2 倍）。這些數據表明，肝臟脂肪含量的大小與肝臟營養過載更為相關——這是一個可以急性改變的因素——而基線高三酸甘油脂血症的部分則只能通過減少能量攝入來緩解。在單變量分析中，胰島素敏感性的改善與來自 DNL 的 VLDL-TG 棕櫚酸比例的降低顯著相關（r = –0.598，P = 0.019；圖 8B）。 在多元回歸分析中，高 LF 組的數據顯示，IHTG 變化是由較高的基線 IHTG（P < 0.0001）和胰島素敏感性指數的較大絕對改善（P = 0.006）所預測的（調整後 R ² = 0.942，F _2,6 = 65.5，P < 0.0001；圖 8C）。在減重後，較高的殘餘 IHTG 是由胰島素敏感性指數的無改善或低改善（P = 0.003）、較低的體重減輕（P = 0.027）以及隨訪時較高的 DNL 水平（P = 0.014）所預測的，減重後的 DNL 是剩餘 IHTG 的最強預測因子（調整後 R ² = 0.849，F _3,5 = 16.0，P = 0.005；圖 8D）。

Peripheral TG metabolism. Finally, weight loss reduced fasting VLDL-TG concentrations (P = 0.040; Figure 7A), particularly in the HighLF group (P = 0.008). This reduction in VLDL-TG concentration was not due to a reduction in VLDL-TG production rate (Figure 7B), but primarily to increased VLDL-TG fractional catabolic rate (P = 0.038; Figure 7C), with a modest increase in clearance rate (P = 0.126 for weight loss; Figure 7D). Across all subjects (Figure 8A), while liver fat was uniformly reduced (as much as 12-fold in some subjects), VLDL-TG concentrations were reduced by a lesser amount (2-fold). These data suggest that the magnitude of liver fat content is more tied to hepatic nutrient overload — a factor that can be acutely modified — while only a proportion of baseline hypertriacylglycerolemia can be mitigated by reducing energy intake. n univariate analysis, the improvements in insulin sensitivity were significantly associated with reductions in the fraction of VLDL-TG palmitate that was derived from DNL (r = –0.598, P = 0.019; Figure 8B). In multiple regression analysis, data from the HighLF group showed that IHTG change was predicted by a higher baseline IHTG (P < 0.0001) and greater absolute improvements in the insulin sensitivity index (P = 0.006) (adjusted R² = 0.942, F_2,6 = 65.5, P < 0.0001; Figure 8C). After weight loss, higher residual IHTG was predicted by no or low improvements in the insulin sensitivity index (P = 0.003), a lower loss of body weight (P = 0.027), and a higher level of DNL at follow-up (P = 0.014), with post–weight loss DNL being the strongest predictor of remaining IHTG (adjusted R² = 0.849, F_3,5 = 16.0, P = 0.005; Figure 8D).

圖 7 Figure 7

VLDL-TG 濃度和週轉率在減重前後的變化。數據來自低脂肪組（LowLF，A 組 n = 6，B–D 組 n = 5，因為有 1 名受試者缺失數據，無法進行動力學分析）和高脂肪組（HighLF，n = 9），顯示減重的影響。(A) 空腹 VLDL-TG 濃度的降低。(B) VLDL-TG 產生率無變化。(C) 分數分解率的增加。(D) 對顆粒-TG 清除率的輕微影響。數據以平均值呈現（在圖表上以“+”表示；可能位於數據點後面）；框代表第 25 和第 75 百分位數；框中的中線代表中位數；而觸鬚代表最小值和最大值。圖表上方的 P 值表示由雙向重複測量方差分析（2-way RM-ANOVA）確定的減重和肝臟脂肪組的主要效應；未觀察到交互作用效應。圖表上的 P 值代表低脂肪組和高脂肪組在基線到隨訪的變化中的子組比較，來自雙向重複測量方差分析（2-way RM-ANOVA）。

VLDL-TG concentrations and turnover rates before and after weight loss. Data are from LowLF subjects (n = 6 for A and n = 5 for B–D because of missing data for 1 subject that precluded kinetic analysis) and HighLF subjects (n = 9) and show impacts of weight loss. (A) Reduction of fasting VLDL-TG concentrations. (B) No changes in VLDL-TG production rates. (C) Increase in fractional catabolic rates. (D) Modest effect on particle-TG clearance rate. Data are presented as mean (indicated by “+” on the graph; may be behind a data point); boxes represent the 25th and 75th percentiles; middle line in the boxes represents the median; and whiskers represent the minimum and maximum values. P values above the graphs indicate main effects of weight loss and liver fat group determined by 2-way RM-ANOVA; no interaction effects were observed. P values on the graphs represent subgroup comparisons from the 2-way RM-ANOVA in the change from baseline to follow-up within LowLF and HighLF groups.

圖 8 Figure 8

肝臟脂肪變化與 VLDL-TG、胰島素敏感性和脂肪生成減少之間的關係。(A) 箭頭顯示高脂肪（HighLF，n = 9）和低脂肪（LowLF，n = 6）受試者肝臟脂肪和血漿 VLDL-TG 濃度的絕對變化幅度。(B) 對於所有受試者，胰島素敏感性（通過靜脈葡萄糖耐受測試評估）的改善越大，空腹 VLDL-TG 棕櫚酸的% DNL 減少越大（通過皮爾森相關性評估）。對於高脂肪受試者，提供了體重減輕後 IHTG 絕對變化的預測因子的逐步回歸係數(C)，以及隨訪時 IHTG 最終水平的預測因子(D)。例如，在 C 中，IHTG 的大幅減少（隨訪減去基線）是由於基線 IHTG 水平較高和胰島素敏感性增加較大所預測的。

Relationships between changes in liver fat and reductions in VLDL-TG, insulin sensitivity, and lipogenesis. (A) Arrows depict the magnitude of absolute changes in liver fat and plasma VLDL-TG concentrations for HighLF (n = 9) and LowLF (n = 6) subjects. (B) For all subjects combined, the greater the improvements in insulin sensitivity (assessed by the insulin sensitivity index via the i.v. glucose tolerance test), the greater the reductions in fasting VLDL-TG palmitate %DNL (assessed by Pearson’s correlation). For the HighLF subjects, stepwise regression coefficients are presented for predictors of the absolute (Abs) change in IHTG with weight loss (C), and for predictors of the final level of IHTG at follow-up (D). For example, in C, large reductions in IHTG (follow-up minus baseline) were predicted by greater baseline IHTG levels and greater increases in insulin sensitivity.

討論 Discussion

本研究使用了一個綜合測試方案，以調查體重減輕對於肝臟脂肪酸和三酸甘油脂代謝的多個方面的影響，研究對象包括肝臟脂肪含量升高（HighLF）和同樣具有胰島素抗性但肝臟脂肪含量較低（LowLF）的受試者。我們評估了 IHTG 的來源、血漿 FFA 流量、全身底物氧化以及 VLDL-TG 動力學，並在禁食和進食狀態以及整個夜間進行了觀察。在禁食狀態下，對於健康的瘦體重受試者，來自脂肪組織儲存的脂肪酸是肝臟用於 VLDL-TG 合成的主要脂肪酸來源，而 DNL 通常僅佔這些來源的約 5%–10%，在禁食狀態下，飲食的貢獻通常較小（23, 24）。在本研究中，對於 HighLF 的受試者，脂肪組織 FFA 的貢獻約為 43%，而來自所有途徑的飲食貢獻約為 4%（其中 1%來自乳糜微粒脂肪酸洩漏進入 FFA 池，3%計算來自乳糜微粒殘餘物的肝臟攝取）。當考慮到直接用於 VLDL-TG 合成的肝臟 DNL（19%）和通過 FFA 池進入肝臟的 DNL 脂肪酸（2%）時，DNL 對 VLDL-TG 棕櫚酸的貢獻約為 21%（圖 6）。 減重後，所有 6 個來源的比例在各組之間相同，特徵是 VLDL-TG 的主要脂肪酸來源來自脂肪組織的脂解（所有受試者合併為 74%），而來自飲食（7%）和去新生脂肪（10%）的量較少。

This study used a comprehensive testing protocol to investigate the impact of weight loss on multiple aspects of fatty acid and triglyceride metabolism in subjects with elevated liver fat (HighLF) and similarly insulin-resistant subjects with low liver fat (LowLF). We assessed the sources of IHTG, plasma FFA flux, whole-body substrate oxidation, and VLDL-TG kinetics, during fasted and fed states and throughout the night. In well-fasted, lean subjects, fatty acids derived from adipose tissue stores are the primary source of fatty acids used by the liver for VLDL-TG synthesis, while DNL typically represents about 5%–10% of these sources, and dietary contribution in the fasting state is usually minor (23, 24). In the present study, for those with HighLF, adipose FFAs contributed about 43%, and diet from all routes contributed about 4% (1% from chylomicron fatty acid spillover into the FFA pool, and 3% calculated to come liver from chylomicron remnant uptake). DNL contributed about 21% to VLDL-TG palmitate when the sum of hepatic DNL directly used for VLDL-TG synthesis (19%) and DNL fatty acids entering the liver through the FFA pool (2%) were accounted for (Figure 6). After weight loss, the proportions of all 6 sources were the same between the groups, characterized by the predominant source of fatty acids for VLDL-TG being derived from adipose lipolysis (74% for all subjects combined) with lower amounts derived from diet (7%) and DNL (10%).

在 VLDL-TG 中棕櫚酸來源的變化及體重減輕的影響。考慮到脂肪組織的游離脂肪酸（FFAs）是肝臟脂質的主要脂肪酸來源，已經有研究建議，血漿 FFAs 流量的增加是 MASLD 中肝臟脂肪積累的主要驅動因素（9）。此外，以前的數據顯示，在胰島素抵抗的狀態下，血漿 FFAs 的夜間出現速率（RaFFA）是升高的（30–32）。然而，在基線時，未觀察到空腹血漿 FFAs 濃度或 FFAs 輸送（RaFFA；圖 5）之間的組別差異，且體重減輕並未導致 RaFFA 的變化，儘管在 HighLF 組中空腹血漿 FFAs 濃度有所降低（補充圖 3C）。相反，體重減輕後 FFAs 對 VLDL-TG 的比例貢獻更大，血漿 FFAs 被回收進入分泌的 VLDL-TG 的速率也更高。這些數據表明，肝臟對 FFAs 的再酯化和儲存的減少是導致 IHTG 減少的一個機制。 考慮到當脂肪酸合成進行時，馬隆酰輔酶 A（malonyl-CoA）會減少 FFA 氧化，因此馬隆酰輔酶 A 的降低很可能會增加肝臟 FFA 氧化以及 FFA 向三酸甘油脂（TG）分泌的路徑。
Alterations in the sources of palmitate in VLDL-TG and effect of weight loss. Given that adipose FFAs are the major source of fatty acids for hepatic lipids, it has been suggested that increased availability of plasma FFA flux is a main driver of liver fat accumulation in MASLD (9). Further, previous data have shown that the nighttime rate of appearance of plasma FFAs (RaFFA) is elevated in states of insulin resistance (30–32). However, at baseline, no group differences were observed in fasting plasma FFA concentrations or FFA delivery (RaFFA; Figure 5), and weight loss did not result in changes in RaFFA, though fasting plasma FFA concentrations were reduced in the HighLF group (Supplemental Figure 3C). By contrast, the proportional contribution of FFAs to VLDL-TG was greater following weight loss, and the rates at which plasma FFAs were recycled into VLDL-TG that was secreted were also greater. These data suggest that a reduction in liver re-esterification and storage of FFAs was one mechanism contributing to the reduction in IHTG. Given that FFA oxidation is reduced by malonyl-CoA (a by-product when fatty acid synthesis is on), it is highly possible that lower malonyl-CoA increased both hepatic FFA oxidation and also routing of FFAs to TG secretion.

DNL 途徑對於過量底物可用性（果糖和葡萄糖碳）以及胰島素抵抗非常敏感，並且可以在數量和質量上顯著促進肝臟脂質的積累（11, 34–36）。我們最近顯示，DNL 的水平在肝病逐漸進展的患者中較高（14）。在數量上，DNL 貢獻了新合成的脂肪酸（11），而在質量上，DNL 途徑的活化也上調了參與脂肪酸酯化和儲存的基因（37）。DNL 在進食狀態下被激活，而在禁食期間則受到抑制，這可以在低脂肪（LowLF）受試者的 VLDL-TG 棕櫚酸比例的變化中觀察到，從午夜（時間 0）到 0500 小時（圖 5G）——而在高脂肪（HighLF）受試者中，這一變化的軌跡則慢得多（圖 5H）。急性能量流對 DNL 速率的重要性體現在這樣的事實上，即這種顯著的體重減輕引起的 DNL 模式變化可以被觀察到，然而這些受試者在研究結束時仍然有肥胖/超重的情況。

The DNL pathway is sensitive to excess substrate availability (fructose and glucose carbons) coupled with insulin resistance and can substantially contribute to hepatic lipid accumulation in both quantitative and qualitative ways (11, 34–36). We have recently shown that the level of DNL is higher in patients with progressively more advanced liver disease (14). Quantitatively, DNL contributes newly made fatty acids (11), and qualitatively, activation of the DNL pathway also upregulates genes involved in esterification and storage of fatty acids from any source (37). DNL is activated in the fed state and becomes suppressed during fasting, which can be observed here in the change in proportion of VLDL-TG palmitate from DNL in the LowLF subjects at baseline from midnight (time 0) to 0500 hours (Figure 5G) — a trajectory that was much slower in the HighLF subjects (Figure 5H). The importance of acute energy flux to the rate of DNL is exemplified by the fact that such marked weight loss–induced changes in DNL patterns can be observed, and yet these subjects still had obesity/overweight at the end of the study.

該干預減少了飲食的總能量含量，特別關注總糖和添加糖。體重減輕導致胰島素敏感性和葡萄糖代謝顯著改善，在高脂肪低碳水化合物組中觀察到空腹狀態下的葡萄糖氧化增加和葡萄糖處置（圖 7B）。在低脂肪組中，較高的外周葡萄糖清除率與較低的進食狀態下的葡萄糖氧化率相結合，支持改善非氧化性葡萄糖處置。在人類代謝研究中，生理變數變化之間的關係可能難以檢測，而如此強勁的發現表明，應該積極倡導增加外周葡萄糖氧化的干預措施，以幫助 MASLD 患者（38-41）。
The intervention reduced overall total energy content of the diet, with specific focus on total and added sugars. Weight loss resulted in notable improvements in insulin sensitivity and glucose metabolism, observed in the HighLF subjects by a greater glucose oxidation in the fasting state and glucose disposal (Figure 7B). In the LowLF group, greater peripheral glucose clearance was coupled with lower fed-state glucose oxidation rates, supporting improved non-oxidative glucose disposal. In human metabolic studies, relationships between changes in physiological variables can be hard to detect, and findings this robust suggest that interventions that increase peripheral glucose oxidation should be vigorously advocated for MASLD patients (38–41).

在進食狀態（在午夜觀察，晚餐後 6 小時）和禁食狀態下，飲食脂肪對 VLDL-TG 棕櫚酸的相對貢獻在 LowLF 和 HighLF 組的基線上是相似的，令人驚訝的是，在減重後的 HighLF 組，肝臟似乎以稍高的速度攝取乳糜微粒殘餘物並將脂質回收至 VLDL（圖 6）。在同位素給予後，VLDL-TG 中未標記的脂肪酸比例在減重後減少了 66%（P = 0.003；圖 6）。我們解釋這一結果是由於用於 VLDL 合成的肝臟-TG 儲存量降低。綜合來看，數據表明，當 DNL 升高時，它會施加額外的脂質負擔——它不會取代其他脂肪酸來源以納入 VLDL-TG，而是通過增加所有脂質來源的重新酯化來增加它們。這一效應導致肝臟脂質過剩和高三酸甘油脂血症。 如果正確，這表明能量限制和體重減輕所引起的肝臟營養過載減少與以下幾點有關：(a) 外周使用葡萄糖，這將有助於降低肝臟可用於去新生脂肪酸合成的底物；(b) 游離脂肪酸被引導遠離肝臟儲存並朝向極低密度脂蛋白分泌；以及 (c) 先前儲存的肝臟三酸甘油脂的損失，推測是通過增加肝臟氧化而耗竭。

The relative contributions of dietary fat to VLDL-TG palmitate in both the fed state (observed at midnight, 6 hours after the dinner meal) and the fasting state were similar between LowLF and HighLF groups at baseline, and surprisingly, in the HighLF group after weight loss, the liver appeared to take up chylomicron remnants and recycle the lipid to VLDL at a slightly higher rate (Figure 6). The proportion of fatty acids in VLDL-TG that remained unlabeled after isotope administration was reduced 66% after weight loss (P = 0.003; Figure 6). We interpret this result to be due to lower availability of hepatic-TG stores used for VLDL synthesis. Taken together, the data suggest that when DNL is elevated it exerts an additional lipid burden — it does not displace other fatty acid sources for inclusion into VLDL-TG but instead adds to them by increasing the re-esterification of all sources of lipid. This effect results in hepatic lipid excess and hypertriacylglycerolemia. If correct, this suggests that energy restriction and weight loss–induced reductions in nutrient overload in the liver are associated with (a) peripheral usage of glucose, which would serve to lower liver substrates available for DNL; (b) FFAs being routed away from hepatic storage and toward VLDL secretion; and (c) loss of previously stored hepatic TG, presumably depleted via increased hepatic oxidation.

RaFFA 的缺乏增加可能看起來是一個悖論，因為一個人減少體脂肪的主要方式似乎是來自脂肪組織被身體燃燒。然而，在 HighLF 組中，我們估計為了產生觀察到的 5.1 ± 0.8 公斤的脂肪質量減少，所需的每日脂肪 RaFFA 增加的幅度為 70 ± 11 μmol/min。這個數量與血漿 FFA 的周轉（約 300–400 μmol/min；圖 5，A 和 B）相比是小的，因此這一變化可能難以檢測。根據 Magkos 等人的計算，我們的樣本大小應該足以檢測到 RaFFA 的這種增加（42）。另一個支持脂肪儲存大小減少而 RaFFA 沒有實質性變化的外周機制是，如果飲食治療減少了脂肪-TG 合成。考慮到減重所產生的較低胰島素濃度，這一機制是可能的，因為這將有助於減少脂肪細胞內的脂肪生成酶和脂質清除到脂肪組織（43）。
The lack of increase in RaFFA may appear a paradox, because it might seem obvious that a principal way a person can lose body fat is if it comes out of the adipose tissue to be burned by the body. However, in the HighLF group, we estimated that the magnitude of the increase in daily adipose RaFFA needed to produce the observed 5.1 ± 0.8 kg reduction in adipose mass cover 6 months was 70 ± 11 μmol/min. This amount is small compared with the turnover of plasma FFAs (~300–400 μmol/min; Figure 5, A and B), and thus this change may be difficult to detect. Using the calculations of Magkos et al., our sample size should have been sufficient to detect such an increase in RaFFA (42). Another peripheral mechanism that would support loss of adipose depot sizes without a substantial change in RaFFA would be if the dietary treatment reduced adipose-TG synthesis. This mechanism was likely, given the lower insulin concentrations produced by weight loss, which would serve to reduce both intra-adipocyte lipogenic enzymes and lipid clearance to adipose tissue (43).

MASLD 中的 VLDL 動力學及減重後的變化。MASLD 患者的心血管代謝風險增加，其特徵為動脈粥樣硬化表型（44, 45）。減重使高脂肪飲食（HighLF）受試者的 HDL 膽固醇提高了 14%（P = 0.045；補充表 1），並使血漿三酸甘油脂（TG）降低了 20%（P = 0.006）及 VLDL-TG 降低了 27%（圖 7A），後者是通過 VLDL-TG 的分數分解率增加 35%（圖 7C）實現的。其他研究者觀察到，經過減重手術後，肥胖個體的 VLDL-TG 分數清除率有所降低（46），或在低能量飲食的腹部肥胖女性中（47）；然而，這兩篇出版物均未包含 IHTG 的測量。在目前的高脂肪飲食組中，IHTG 的減少幅度與 VLDL-TG 濃度的減少相比較大（圖 8A）。這突顯了 IHTG 受到急性能量限制的強烈驅動（48），而血漿脂質濃度可能受到更多的遺傳控制（49, 50）。我們的數據表明，外周適應以減少肝臟壓力的能力有助於防止 MASLD。 在進食狀態下，晚餐後觀察到需要較低的胰島素水平來維持葡萄糖濃度，顯示出更好的底物處理（39, 51, 52）。在這種情況下，將營養物質重新導向遠離肝臟，朝向外周儲存和氧化，減少了肝臟的毒性。

VLDL kinetics in MASLD and after weight loss. The increased cardiometabolic risk of patients with MASLD is characterized by an atherogenic phenotype (44, 45). Weight loss raised the HighLF subjects’ HDL-cholesterol by 14% (P = 0.045; Supplemental Table 1) and lowered plasma TG by 20% (P = 0.006) and VLDL-TG by 27% (Figure 7A), the latter occurring via a 35% increase in VLDL-TG fractional catabolic rates (Figure 7C). Other investigators have observed a reduction in VLDL-TG fractional clearance rate in individuals with obesity following weight loss via gastric bypass surgery (46) or in women with abdominal obesity on a low-energy diet (47); however, neither publication included measures of IHTG. In the present HighLF group, the magnitude of reductions in IHTG was large in comparison with reductions in VLDL-TG concentrations (Figure 8A). This highlights the fact that IHTG is strongly driven by acute energy restriction (48) whereas plasma lipid concentrations may be more genetically controlled (49, 50). Our data suggest that the ability of the periphery to adapt in a manner that reduces the stress on the liver protects against MASLD. In the fed state, better substrate disposal was evident after dinner in the observed lower levels of insulin needed to maintain glucose concentrations (39, 51, 52). In this scenario, rerouting of nutrients away from the liver, toward peripheral storage and oxidation, reduces hepatic toxicity.

限制與未來研究需求。本研究的主要限制與使用胰島素敏感性和同位素標記的調查性質有關，這需要密集的方案，通常會限制研究樣本的大小。血漿游離脂肪酸（FFAs）和去新生脂肪酸（DNL）的標記與文獻中的其他工作不同，後者主要集中於分別研究這兩個來源（25, 46, 47, 53）。測量空腹 VLDL-TG 產生速率以評估穩態下的合成；如果在進食狀態下評估，VLDL-TG 的來源可能會不同。棕櫚酸被用作所有脂肪酸的標記，因為它是 DNL 的主要產物，是飲食中第二最突出的脂肪酸，也是血漿 FFAs 和 VLDL-TG 中的主要脂肪酸之一。DNL 對其他脂肪酸的貢獻是變化的，並且不會對亞油酸等必需脂肪酸作出貢獻。因此，DNL 對總脂肪酸的貢獻難以估計，並且可能低於棕櫚酸的觀察值。 在飲食介入研究中，需要許多工作人員和資源來支持受試者並限制長期（6 個月治療）中的脫落率，而這個項目還有額外的受試者負擔，即測量夜間流量率。兩組的體重減輕軌跡在 6 個月的調查中保持幾乎線性，這有些不尋常，而不是在開始時急劇下降並在結束時放緩。持續的體重減輕率是因為營養師在整個研究中強調逐漸減少食物攝入。最後 3 週在減重後測試期間，體重保持穩定，以在隨訪入院時達到一致的代謝狀態。儘管比某些介入研究的時間更長，但 6 個月的減重時間可能不足以使所有受試者的 IHTG 水平降至 5.5%以下。

Limitations and future research needs. The primary limitation of this study relates to the nature of investigations using measures of insulin sensitivity and isotopic labeling, which require intensive protocols that typically limit study sample sizes. Labeling of plasma FFAs and DNL is distinct from other work in the literature, which has had a focus on investigating either of these 2 sources separately (25, 46, 47, 53). Fasting VLDL-TG production rate was measured to assess synthesis during a steady state; the sources of VLDL-TG would likely be different if assessed in the fed state. Palmitate was used as the marker for all fatty acids because it is the primary product of DNL, the second most prominent fatty acid in the diet, and one of the main fatty acids in plasma FFAs and VLDL-TG. The contribution of DNL to other fatty acids is variable, and it would not contribute to essential fatty acids like linoleic acid. Therefore, DNL contribution to total fatty acids is challenging to estimate and is likely lower than what is observed for palmitate. For dietary intervention studies, many staff and resources were needed to support subjects and limit dropout rates over long durations (6 months of treatment), and this project had the additional subject burden of measuring nocturnal flux rates. It is slightly unusual that the weight loss trajectories of both groups remained nearly linear over the 6-month investigation, rather than falling precipitously at the beginning and slowing toward the end. The continued rate of weight loss was due to the dietitian emphasizing progressively greater reductions in food intake throughout the study. The final 3 weeks during post–weight loss testing, body weight was held steady to achieve a consistent metabolic state at the follow-up admission. Although longer than some intervention studies, the duration of 6 months for weight loss may not have been long enough to cause all subjects to reach IHTG levels below 5.5%.

VLDL-TG 棕櫚酸來源的計算基於補充方法中描述的一些模型假設。用同位素測量的 TG 代謝的一個特徵是，在數據獲取過程中，標籤不可避免地會有一些回收。這在血漿 FFA 池中 DNL 脂肪酸的存在中顯而易見，這些脂肪酸可能來自於通過重水給藥標記的脂肪組織脂肪酸的釋放，或者可能來自 VLDL-DNL 脂肪酸的溢出。考慮到 10 天的 D 2 O 給藥時間相對較短，無法檢測到脂肪組織 TG 的標記，我們更傾向於後者解釋 FFAs 中 DNL 的存在。來自肝臟 DNL 和 FFA-DNL 的絕對 VLDL-TG 濃度呈正相關（基線時 r = 0.852，P = 0.001；隨訪時 r = 0.884，P = 0.002）。模型中的第二個未知數是 VLDL-TG 中未標記脂質的來源。這可能是由於在進入肝臟時棕櫚酸標籤的未測量內臟稀釋，或來自於緩慢轉換的肝內脂質小滴，這些小滴為 VLDL-TG 合成貢獻了未標記的脂質。 我們對 MASLD 患者脂質標記的原始研究顯示，儲存脂質的量與 VLDL-TG 中的未標記脂質之間存在強烈的關聯（12）。考慮到本研究的減重干預，IHTG 和內臟脂肪可能都減少了，因此，未標記池的真正來源無法明確識別。第三，該研究並未設計用於了解保護 LowLF 組在基線時免受脂肪肝影響的獨特機制。考慮到這些受試者的肥胖和胰島素抵抗水平相似，了解這種保護將是重要的。應該調查瘦型 MASLD（54）中的脂肪生成測量，以確定該人群中的 DNL 是否升高。我們相信，胰島素抵抗和遺傳因素，而不僅僅是過多的體重，與 MASLD 有關。最後，儘管這裡研究的 HighLF 受試者在 IHTG 上顯示出顯著減少，但他們減重後的 IHTG 值仍然高於 LowLF 組。 儘管已經確定了 IHTG 的遺傳因素（55），但迄今為止尚無研究測試家庭內脂肪生成的遺傳性。生活方式因素（例如，飲食攝取水平、運動）可以影響 IHTG 的水平，而本研究專注於能量限制以降低肝臟營養過載。還需要進一步的研究來揭示通過運動增加能量消耗如何也可能限制營養毒性（38, 56, 57）並改善能量限制的效果（58）。

The calculation of sources of VLDL-TG palmitate is based on a number of model assumptions described in Supplemental Methods. One characteristic of TG metabolism measured with isotopes is that it invariably results in some recycling of labels during data acquisition. This was evident by the presence of DNL fatty acids in the plasma FFA pool, which could have been derived from release of adipose fatty acids that were labeled by administration of heavy water or could have been derived from spillover of VLDL-DNL fatty acids. Given that 10 days is a relatively short duration of D2O administration to be able to detect labeling of adipose TG, we favor the latter explanation for the presence of DNL in FFAs. Absolute VLDL-TG concentrations from hepatic DNL and FFA-DNL were positively correlated (r = 0.852, P = 0.001 at baseline and r = 0.884, P = 0.002 at follow-up). A second unknown in the model is the source of unlabeled lipid in VLDL-TG. It could have resulted from unmeasured visceral dilution of the palmitate label as it entered the liver or from intrahepatic lipid droplets that slowly turned over, contributing unlabeled lipid to VLDL-TG synthesis. Our original study of lipid labeling in MASLD patients demonstrated a strong association between the magnitude of stored lipid and the unlabeled lipid in VLDL-TG (12). Given the present study’s weight loss intervention, both IHTG and visceral fat were likely reduced, and thus, the true source of the unlabeled pool cannot be definitively identified. Third, the study was not designed to understand the unique mechanisms that protected the LowLF group from fatty liver at baseline. Given that these subjects had similar levels of obesity and insulin resistance, understanding this protection would be important to accomplish. Measurement of lipogenesis in lean MASLD (54) should also be investigated to determine whether DNL is elevated in this population. It is our belief that both insulin resistance and genetic factors, rather than just excess body weight, are associated with MASLD. Lastly, although the HighLF subjects investigated here exhibited a considerable reduction in IHTG, their post–weight loss IHTG values were still greater than those of the LowLF group. Although genetic contributors to IHTG have been identified (55), no studies to date have tested the heritability of lipogenesis within families. Lifestyle factors (e.g., levels of dietary intake, exercise) can influence levels of IHTG, and the present study focused on energy restriction to lower hepatic nutrient overload. Additional work is needed to uncover how increases in energy expenditure through exercise may limit nutrient toxicity also (38, 56, 57) and improve the effect of energy restriction (58).

結論。本研究的發現顯著推進了對 MASLD 發展中脂肪酸代謝的整體理解。當量化所有脂肪酸來源對 IHTG 的貢獻時，過量的脂肪生成是肝臟脂肪的主要驅動因素。胰島素抵抗，先前已顯示與過量 IHTG 有關，隨著體重減輕而顯著改善，這些改善預測了肝臟脂肪的減少。體重減輕還顯著改善了在晚餐後和整個夜間延長禁食時觀察到的基線緩慢晝夜變化。在我們之前針對胰島素抵抗匹配的受試者的基線研究中，發現升高的 DNL 是區分高脂肪肝（HighLF）和低脂肪肝（LowLF）受試者的唯一變數（11）。在這裡，我們顯示當通過能量限制顯著減少肝臟脂肪時，脂肪生成的減少在很大程度上解釋了肝臟的改善，影響了肝臟對其他來源脂肪酸的處理。在干預結束時，當過量 IHTG 得到改善時，脂肪生成水平的個體差異仍然是殘餘 IHTG 的驅動因素。 因為去新脂肪合成（DNL）對能量攝取的減少反應非常敏感，這條途徑代表了對於 MASLD 來說最具可調整性的生化貢獻者。目前，減少肝臟脂質合成是這個領域藥物開發的目標。在此同時，應更強調幫助患者減重作為治療策略的努力。

Conclusions. The present findings substantially advance the integrated understanding of fatty acid metabolism in the development of MASLD. When the contribution of all sources of fatty acids to IHTG was quantitated, it was excess lipogenesis that was the primary driver of liver fat. Insulin resistance, previously shown to be implicated in excess IHTG, was markedly improved with weight loss, and these improvements predicted loss of liver fat. Weight loss also substantially improved the observed baseline slow circadian transitions found after dinner and through the night as fasting was extended. In our previous baseline studies of subjects matched for insulin resistance, elevated DNL was found to be the singular variable segregating HighLF and LowLF subjects (11). Here, we show that when liver fat is substantially reduced by energy restriction, it is the reduction in lipogenesis that largely accounted for the liver improvements, influencing liver processing of fatty acids from other sources. At the end of the intervention when excess IHTG was ameliorated, individual differences in lipogenesis levels remained the driver of residual IHTG. Because DNL is highly responsive to reductions in energy intake, this pathway represents the most highly modifiable biochemical contributor to MASLD. Reductions in hepatic lipid synthesis are currently a target for pharmaceutical development in this field. In the meantime, efforts to help patients lose weight should be more strongly emphasized as a strategy for treatment.

方法 Methods

性別作為生物變數。我們的研究包括男性和女性。我們按性別分析了主要結果（體重變化、肝臟脂肪、脂肪酸來源）的數據，結果在男性和女性之間沒有差異（數據未顯示），因此結果以群體形式呈現。

Sex as a biological variable. Our study included men and women. We analyzed the data for the primary outcomes (changes in body weight, liver fat, fatty acid sources) by sex, and results were not different between men and women (data not shown), and so results are presented as a group.

受試者。這些方法大部分已在之前的文獻中發表（11, 59, 60），在此及補充方法中簡要描述。從當地社區健康博覽會和醫生轉介中招募了 16 名非糖尿病、非吸煙的西班牙裔或非裔美國人（自我描述）超重/肥胖受試者，這些受試者的體重穩定，使用了兩部分的篩選協議和特定標準，以提高找到 MASLD 受試者的可能性（補充圖 1 和補充方法）。受試者在基線時根據肝臟 TG 含量分為低（LowLF）或高（HighLF）肝脂肪組，分別為<5.6%或≥5.6%，測量方法為 ¹ H-MRS（1）。本次分析的目標是確定通過能量限制所產生的體重減輕如何影響這些受試者的肝臟脂肪酸合成、TG 組裝和外周代謝。在參與干預的 16 名受試者中，所有 16 名均完成了干預，但因為低遵從性（缺席約診），高肝脂肪組中的 1 名受試者被排除在分析之外。

Subjects. The methods have largely been published previously (11, 59, 60) and are described briefly here and in Supplemental Methods. Sixteen non-diabetic, non-smoking Hispanic or African American (self-described) overweight/obese subjects with stable body weight were recruited from local community health fairs and physician referral using a 2-part screening protocol and specific criteria to increase the likelihood of finding subjects with MASLD (Supplemental Figure 1 and Supplemental Methods). Subjects were stratified at baseline into either low (LowLF) or high (HighLF) liver fat groups based on hepatic TG content of <5.6% or ≥5.6%, respectively, as measured by ¹H-MRS (1). The goal of the present analysis was to determine how weight loss occurring through energy restriction impacted hepatic fatty acid synthesis, TG assembly, and peripheral metabolism in these subjects. Of the 16 subjects who participated in the intervention, all 16 completed the intervention, but 1 subject in the HighLF group was excluded from analysis because of low adherence (missing appointments).

研究設計。如圖 1A 所示，本研究包括三個階段：(a) 基線研究，(b) 約 5-6 個月的減重介入期，以及(c) 在達到體重穩定後的隨訪研究（約在 6 個月時進行）。基線和隨訪研究在 3 週的測試期間進行，並包括在德克薩斯州大學西南醫學中心臨床與轉化研究中心的兩次住院過夜。第一次（住院 1）旨在通過頻繁取樣的胰島素修飾靜脈葡萄糖耐受測試來測量胰島素敏感性（60），第二次（住院 2）則旨在測量脂肪酸代謝（圖 1B）。受試者在測試期間保持身體活動水平，所有食物和飲料均根據他們的習慣飲食模式（如下所述）在兩次住院前和期間提供給受試者。

Study design. As shown in Figure 1A, this investigation comprised 3 stages: (a) baseline studies, (b) a weight loss intervention period of approximately 5–6 months, and (c) follow-up studies after weight stability was achieved (conducted at approximately the 6-month mark). Baseline and follow-up studies occurred over 3 weeks of testing and consisted of 2 inpatient overnight admissions in the University of Texas Southwestern Medical Center Clinical and Translational Research Center. The first (admission 1) was designed to measure insulin sensitivity using a frequently sampled, insulin-modified, intravenous glucose tolerance test (60), and the second (admission 2) was designed to measure fatty acid metabolism (Figure 1B). Subjects maintained physical activity levels during the testing period, and all foods and beverages were provided to the subjects before and during both admissions based on their habitual dietary patterns (described below).

程序。基線和隨訪時進行的程序相同，如圖 1B 所示。在入院 1 之前，受試者完成了為期 3 天的食物攝取日記、24 小時飲食回顧，以及與註冊營養師的深入訪談，以確定通常的飲食攝取。食物記錄使用營養數據系統進行分析（NDSR 2009；美國明尼阿波利斯），並根據這些信息，營養師制定了一個維持體重的菜單，以類似受試者的通常食物攝取。食物由臨床與轉化研究中心的廚房準備，並送到受試者家中，供其在 10 天內食用（入院 1 前 3 天和入院 2 前 7 天）。在入院 1 前 3 天及整個入院 2 期間禁止飲酒。受試者在 2 週測試期間保持穩定的體重，並維持入學前的身體活動水平。

Procedures. Procedures performed at baseline and follow-up were identical and are described in Figure 1B. Prior to admission 1, subjects completed a 3-day food intake diary, 24-hour dietary recall, and an in-depth interview with a registered dietitian to determine usual dietary intake. Food records were analyzed using the Nutrition Data System for Research (NDSR 2009; Minneapolis, Minnesota, USA), and using this information, a weight-maintaining menu was formulated by the dietitian to resemble the subjects’ usual food intake. Food was prepared by the Clinical and Translational Research Center kitchen and delivered to the subject to consume at home for 10 days (3 days before admission 1 and 7 days before admission 2). Alcohol consumption was prohibited from 3 days before admission 1 and throughout the period through admission 2. Subjects had stable body weight and maintained pre-enrollment physical activity levels before and during the 2-week testing period.

在進行靜脈注射葡萄糖耐受測試以評估葡萄糖和胰島素反應及胰島素抗性（補充方法及參考文獻 60）當天，還通過 DEXA 掃描儀（Hologic Discovery W，QDR 系列）進行了身體組成測量，並通過 ¹ H-MRS 進行了 IHTG 測量。

On the day of the i.v. glucose tolerance test for assessment of glucose and insulin responses and insulin resistance (Supplemental Methods and ref. 60), body composition measurements were also made by a DEXA scanner (Hologic Discovery W, QDR series), and IHTG measurements by ¹H-MRS.

TG–脂肪酸組成及其對血漿 VLDL-TG 的貢獻來源已顯示反映肝臟 TG（12, 14, 61），因此 VLDL-TG 的特徵用於評估肝臟 TG 的流量。用於追蹤 VLDL-TG 棕櫚酸中所有脂肪酸來源的多重穩定同位素程序來自 DNL、餐飲脂肪和血漿 FFA，詳情見補充方法，包括氘標記水（D ₂ O）的加載和維持劑量、包含[U- ¹³ C ₁₆ ]棕櫚酸的晚餐，以及從午夜開始的靜脈注射輸注，包含[1,2,3,4- ¹³ C ₄ ]棕櫚酸。在第二天中午，餵食與晚餐相同組成和能量的餐。呼吸氣體分析通過間接熱量計（Vmax Encore 代謝車，Viasys Healthcare）在第二天 0830 小時進行，處於禁食狀態，然後在 1430 小時午餐後再次進行，以評估進食的代謝反應。能量消耗及脂肪和葡萄糖氧化使用標準方程計算（62）；蛋白質分解基於受試者前一天的控制蛋白質攝入量計算（根據 NDSR 飲食分析計算）（11）。

The TG–fatty acid composition and sources contributing to plasma VLDL-TG have been shown to reflect liver-TG (12, 14, 61), and therefore the characteristics of VLDL-TG are used to assess liver-TG fluxes. The multiple-stable-isotope procedure used to trace all sources of fatty acids in VLDL-TG palmitate from DNL, meal fat, and plasma FFAs is described in Supplemental Methods, including loading and maintenance doses of deuterium-labeled water (D₂O), the evening meal containing [U-¹³C₁₆]palmitate, and i.v. infusion starting at midnight containing [1,2,3,4-¹³C₄]palmitate. At noon on day 2, a meal of identical composition and energy to the evening meal was fed. Respiratory gas analysis was conducted via indirect calorimetry (Vmax Encore metabolic cart, Viasys Healthcare) at 0830 hours on day 2 in the fasted state and then again after lunch at 1430 hours to assess the metabolic response to eating. Energy expenditure and fat and glucose oxidation were calculated using standard equations (62); protein catabolism was calculated based on the subject’s controlled protein intake from the previous day (calculated from NDSR dietary analysis) (11).

飲食干預。干預的主要目標是通過減少能量攝入和改變食物質量及組成來實現體重減輕。綜合飲食干預方法在補充方法中有進一步描述。

Dietary intervention. The primary goal of the intervention was to achieve weight loss through reduction in energy intake and modification of food quality and composition. The comprehensive dietary intervention methods are further described in Supplemental Methods.

結果。主要結果測量包括體重減輕對肝臟脂肪的變化、脂肪酸來源在肝臟脂質棕櫚酸生產中的利用，以及反映肝臟脂質處理的脂蛋白動力學。次要結果包括體重和組成的變化，以及在禁食和進食狀態下的代謝反應（包括胰島素敏感性和游離脂肪酸動力學）。附加結果的數據（例如，腰圍、LDL-膽固醇、HDL-膽固醇的變化）包含在補充表 1 中。

Outcomes. Primary outcome measurements included changes induced by weight loss in liver fat, utilization of sources of fatty acids for hepatic lipid palmitate production, and lipoprotein kinetics reflecting hepatic handling of lipids. Secondary outcomes included changes in body weight and composition and metabolic responses in the fasted and fed states (including insulin sensitivity and FFA kinetics). Additional data from ancillary outcomes (e.g., changes in waist circumference, LDL-cholesterol, HDL-cholesterol) are included in Supplemental Table 1.

實驗室分析。從入院第 1 天的 1800 時到第 2 天的 1800 時，共在 34 個時間點收集血液，並立即分離血漿以測量 FFA、TG、葡萄糖和胰島素濃度（補充方法和參考文獻 24）。為了分析體重減輕的變化，空腹值是通過平均來自篩查抽血、靜脈葡萄糖耐受測試前抽血以及入院第 2 天 0600 至 0800 時間點的數據來定義的。在隨訪測試期間，對於 1 名受試者，靜脈葡萄糖耐受測試（入院 1）期間的空腹胰島素值超過平均值的 2 個標準差，但在餐後測試（入院 2）期間的空腹胰島素值則在所有受試者的標準差範圍內，並與其他隨體重減輕而改善的代謝指標一致（即肝脂肪顯著減少、DNL、天冬氨酸轉氨酶和丙氨酸轉氨酶的顯著降低）；因此，對於這名受試者，在隨訪中使用的唯一空腹胰島素值是餐後測試期間獲得的空腹胰島素值。

Laboratory analyses. From 1800 on day 1 through 1800 on day 2 of admission 2, blood was collected at 34 time points and plasma immediately separated for measurement of FFA, TG, glucose, and insulin concentrations (Supplemental Methods and ref. 24). For analysis of change with weight loss, fasting values were defined by averaging data from a combination of screening blood draws, those taken just before the intravenous glucose tolerance test, and from admission 2 time points 0600 to 0800. For 1 subject at the follow-up testing period, the fasting insulin value during the i.v. glucose tolerance test (admission 1) was more than 2 SD above the average, but the fasting insulin values during the meal test (admission 2) fell within the SD of all subjects and in line with other metabolic indicators that improved with weight loss (i.e., significant reduction in liver fat, DNL, aspartate transaminase, and alanine transaminase); therefore, for this subject the fasting insulin obtained during the meal test was the only fasting insulin value used at follow-up.

總三酸甘油脂豐富的脂蛋白（TRLs）在 22 個時間點（從午夜到中午，然後在午餐後）從血漿中分離出來（詳見補充方法）。在午夜，TRL 顆粒包含來自前一晚餐的乳糜微粒和肝臟來源的極低密度脂蛋白（VLDL）顆粒的混合物（63）。隨著禁食的進行，乳糜微粒從血漿中被清除，因此在最後一餐後 18 小時分離的 TRL 含有非常低濃度的乳糜微粒。因此，午夜及之後分離的這一部分在此稱為 TRL（包含乳糜微粒和 VLDL），而在長時間禁食後（最後一餐後 18 小時），這一部分稱為 VLDL（33，64）。禁食時間點的平均值（午餐前）被認為反映了 VLDL-TG 棕櫚酸的禁食脂肪酸來源。午餐後的脂肪酸來源數據是為了讀者的興趣而包含，但因為這不是一個穩態期間，這些數據未進行統計評估。 從 TRL 和 TG–脂肪酸分離的 TGs，以及來自血漿的 FFAs，已準備好進行氣相色譜–質譜分析（補充方法和參考文獻 23、24、65–66）。

Total TG-rich lipoproteins (TRLs) were isolated from plasma at 22 time points (from midnight until noon, and then also following the lunch meal; described in Supplemental Methods). At midnight, TRL particles contain a mixture of chylomicrons (from the previous evening meal) and hepatically derived VLDL particles (63). With progressive fasting, chylomicrons are cleared from the plasma such that TRL isolated 18 hours after the last meal contains very low concentrations of chylomicrons. Accordingly, this fraction isolated at and after midnight is referred to herein as TRL (containing both chylomicrons and VLDL), whereas after an extended fast (18 hours after the last meal), this fraction is referred to as VLDL (33, 64). An average of the fasting time points (before the lunch meal) was considered to reflect the fasting fatty acid sources for VLDL-TG palmitate. Data for fatty acid sources after the lunch meal were included for the reader’s interest, but because it is not a steady-state period, these data were not assessed statistically. TGs separated from TRL and TG–fatty acids, as well as FFAs from plasma, were prepared for gas chromatography–mass spectrometry (Supplemental Methods and refs. 23, 24, 65–66).

脂質代謝的量化。貢獻於脂蛋白-TG 棕櫚酸生產的潛在脂肪酸來源（膳食脂肪、脂肪組織脂肪酸和肝臟 DNL）使用已建立的多重穩定同位素程序（23, 24, 33, 65, 66）標記為不同的棕櫚酸同位素。多重穩定同位素技術的詳細信息在補充方法和參考文獻 67 中描述。使用了六種策略來提高同位素標記方法的嚴謹性，包括對 RaFFA 計算的修正（a）同位素純度，（b）乳糜微粒溢出（24），（c）攜帶在注入白蛋白上的未標記脂肪酸，以及（d）棕櫚酸對血漿 FFA 組成的相對貢獻，還有對 VLDL-TG 棕櫚酸合成的計算，（e）血漿 FFA 池中脂原性脂肪酸的數量，以及（f）可以回收進入 VLDL 的血漿 FFA 池中的膳食脂肪酸。

Quantification of lipid metabolism. The potential fatty acid sources that contribute to lipoprotein-TG palmitate production (dietary fat, adipose fatty acids, and hepatic DNL) were each labeled with a different palmitate isotope using an established multiple-stable-isotope procedure (23, 24, 33, 65, 66). Details on the multiple-stable-isotope technique are described in Supplemental Methods and ref. 67. Six strategies were used to increase rigor of the isotope labeling methods, including corrections for the calculation of RaFFA for (a) isotopic purity, (b) chylomicron spillover (24), (c) unlabeled fatty acid carried on infused albumin, and (d) the relative contribution of palmitate to plasma FFA composition, and for calculations for VLDL-TG palmitate synthesis, (e) the quantity of lipogenic fatty acids in the plasma FFA pool, and (f) dietary fatty acids in the plasma FFA pool that could recycle into VLDL.

統計學。計算使用 Excel（版本 2007，微軟）進行。高脂肪和低脂肪組別因體重減輕而從基線到隨訪的變化，使用雙向重複測量方差分析（RM-ANOVA）進行比較，以評估（a）體重減輕和（b）肝臟脂肪組別的影響（使用 Statview for Windows v5.0.1，SAS Institute；或 GraphPad Prism v10）。在住院代謝測試期間的時間過程結果中，我們使用三向 RM-ANOVA（或混合效應模型如果有缺失值）來評估（a）住院測試期間的時間，（b）體重減輕，以及（c）肝臟脂肪組別的影響（GraphPad）。配對或未配對的雙尾 t 檢驗由 Excel（版本 2007 和 365，微軟）進行。多重比較的分析包括 Šidák（對於雙向 RM-ANOVA）或 Tukey（對於三向 RM-ANOVA）修正。變數之間的關係由 Pearson 相關性（GraphPad 或 Excel）確定。 逐步回歸分析用於識別哪些基線變數最佳預測體重減輕對肝臟脂肪的變化；顯著變數被納入模型以評估模型的預測能力，報告為調整後的 R ² （SPSS v21，IBM 軟體）。P 值小於 0.05 被視為統計上顯著，P < 0.10 被報告為趨勢。P 值未針對多重比較進行調整，但多重比較僅限於主要結果。

Statistics. Calculations were performed using Excel (version 2007, Microsoft). Changes from baseline to follow-up due to weight loss in the HighLF and LowLF groups were compared with 2-way repeated-measures ANOVA (RM-ANOVA) for singular outcomes to assess effects of (a) weight loss and (b) liver fat group (using Statview for Windows v5.0.1, SAS Institute; or GraphPad Prism v10). For time course outcomes during the inpatient metabolic tests, we used 3-way RM-ANOVA (or mixed-effects model if missing values) to assess effects of (a) time over the inpatient testing period, (b) weight loss, and (c) liver fat group (GraphPad). Paired or unpaired 2-tailed t tests were performed by Excel (versions 2007 and 365, Microsoft). Analyses with multiple comparisons included Šidák’s (for 2-way RM-ANOVA) or Tukey’s (for 3-way RM-ANOVA) corrections. Relationships between variables were determined by Pearson’s correlation (GraphPad or Excel). Stepwise regression was used to identify which baseline variables best predicted change in liver fat with weight loss; significant variables were entered into a model to assess the predictive power of the model, reported as the adjusted R² (SPSS v21, IBM Software). P values less than 0.05 were considered statistically significant, with P < 0.10 reported as a trend. P values were not adjusted for multiple comparisons, but multiple comparisons were limited to the primary outcomes.

研究批准。該研究已獲得德克薩斯大學西南醫學中心的機構審查委員會批准（IRB 062007-025），並已從所有受試者獲得書面知情同意。

Study approval. The study was approved by the Institutional Review Board at the University of Texas Southwestern Medical Center (IRB 062007-025), and written informed consent was obtained from all subjects.

參考文獻 References

1. 1. Browning JD 等人。美國城市人口肝脂肪變性流行率：種族影響。肝臟病學。2004;40(6):1387–1395。
      Browning JD, et al. Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity. Hepatology. 2004;40(6):1387–1395.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 2. Kanwal F 等。與代謝功能障礙相關的脂肪肝病：更新及新命名對美國肝臟病學會非酒精性脂肪肝病實踐指導的影響。肝臟病學。2024;79(5):1212–1219。
      Kanwal F, et al. Metabolic dysfunction-associated steatotic liver disease: update and impact of new nomenclature on the American Association for the Study of Liver Diseases practice guidance on nonalcoholic fatty liver disease. Hepatology. 2024;79(5):1212–1219.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 3. Rinella ME 等。關於新脂肪肝病命名的多社會德爾菲共識聲明。肝臟學雜誌。2023;79(6):1542–1556。
      Rinella ME, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol. 2023;79(6):1542–1556.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 4. Younossi ZM 等。美國過去三十年慢性肝病的流行病學。Gut。2020；69(3)：564–568。
      Younossi ZM, et al. Epidemiology of chronic liver diseases in the USA in the past three decades. Gut. 2020;69(3):564–568.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 5. Younossi ZM 等。2013-2022 年美國成人肝臟移植流行病學的變化：代謝功能障礙相關脂肪肝病和酒精相關肝病的主導地位。Hepatol Commun。2024；8(1)：e0352。
      Younossi ZM, et al. The changing epidemiology of adult liver transplantation in the United States in 2013-2022: the dominance of metabolic dysfunction-associated steatotic liver disease and alcohol-associated liver disease. Hepatol Commun. 2024;8(1):e0352.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 6. Lee BP 等。使用共識命名法的脂肪肝疾病及其子分類的全國流行病學估計。肝臟病學。2024;79(3):666–673。
      Lee BP, et al. National prevalence estimates for steatotic liver disease and subclassifications using consensus nomenclature. Hepatology. 2024;79(3):666–673.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 7. Zheng X 等。美國青少年中代謝功能障礙相關脂肪肝疾病的流行率及心代謝風險因素。[2024 年 8 月 13 日在線發表]。臨床內分泌與代謝雜誌。https://doi.org/10.1210/clinem/dgae553。
      Zheng X, et al. Prevalence of metabolic dysfunction-associated steatotic liver disease and cardiometabolic risk factor in US adolescents. [published online August 13, 2024]. J Clin Endocrinol Metab. https://doi.org/10.1210/clinem/dgae553.
      通過以下方式查看本文：PubMed Google Scholar
      View this article via: PubMed Google Scholar

1. 8. Iozzo P 等。糖耐量受損患者肝臟對游離脂肪酸的處理缺陷。J Clin Endocrinol Metab。2004;89(7):3496–3502。
      Iozzo P, et al. Defective liver disposal of free fatty acids in patients with impaired glucose tolerance. J Clin Endocrinol Metab. 2004;89(7):3496–3502.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 9. Bril F 等。非酒精性脂肪肝病中肝內三酸甘油脂含量的代謝和組織學意義。Hepatology。2017;65(4):1132–1144。
      Bril F, et al. Metabolic and histological implications of intrahepatic triglyceride content in nonalcoholic fatty liver disease. Hepatology. 2017;65(4):1132–1144.
      通過以下方式查看本文：CrossRef PubMed Google Scholar
      View this article via: CrossRef PubMed Google Scholar

1. 10. Sanyal AJ 等。非酒精性脂肪性肝炎：胰島素抵抗與線粒體異常的關聯。胃腸病學。2001；120(5)：1183–1192。
       Sanyal AJ, et al. Nonalcoholic steatohepatitis: association of insulin resistance and mitochondrial abnormalities. Gastroenterology. 2001;120(5):1183–1192.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 11. Lambert JE 等。新生脂肪生成的增加是非酒精性脂肪肝病患者的一個明顯特徵。胃腸病學。2014；146(3)：726–735。
       Lambert JE, et al. Increased de novo lipogenesis is a distinct characteristic of individuals with nonalcoholic fatty liver disease. Gastroenterology. 2014;146(3):726–735.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 12. Donnelly KL 等。非酒精性脂肪肝疾病患者肝臟中儲存的脂肪酸來源及通過脂蛋白分泌的脂肪酸。J Clin Invest。2005；115(5)：1343–1351。
       Donnelly KL, et al. Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease. J Clin Invest. 2005;115(5):1343–1351.
       透過以下方式查看本文：JCI CrossRef PubMed Google Scholar
       View this article via: JCI CrossRef PubMed Google Scholar

1. 13. Diraison F 等。肝臟新生脂肪生成及血漿非酯化脂肪酸的再酯化對非酒精性脂肪肝疾病期間血漿三酸甘油脂合成的貢獻。Diabetes Metab。2003；29(5)：478–485。
       Diraison F, et al. Contribution of hepatic de novo lipogenesis and reesterification of plasma non esterified fatty acids to plasma triglyceride synthesis during non-alcoholic fatty liver disease. Diabetes Metab. 2003;29(5):478–485.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 14. Syed-Abdul MM 等人。對不同組織學分級肝病患者的肝臘三酸甘油酯進行同位素標記和生化評估。J Nutr。2023；153(12)：3418–3429。
       Syed-Abdul MM, et al. Isotope labeling and biochemical assessment of liver-triacylglycerol in patients with different levels of histologically-graded liver disease. J Nutr. 2023;153(12):3418–3429.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 15. Lawitz EJ 等人。乙酰輔酶 A 羧化酶抑制劑 GS-0976 在 12 週內減少非酒精性脂肪肝炎患者的肝臟新生脂肪生成和脂肪變性。Clin Gastroenterol Hepatol。2018；16(12)：1983–1991。
       Lawitz EJ, et al. Acetyl-CoA carboxylase inhibitor GS-0976 for 12 weeks reduces hepatic de novo lipogenesis and steatosis in patients with nonalcoholic steatohepatitis. Clin Gastroenterol Hepatol. 2018;16(12):1983–1991.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 16. Loomba R 等。TVB-2640（FASN 抑制劑）用於治療非酒精性脂肪肝炎：FASCINATE-1，一項隨機、安慰劑對照的 2a 階段試驗。胃腸病學。2021；161(5)：1475–1486。
       Loomba R, et al. TVB-2640 (FASN inhibitor) for the treatment of nonalcoholic steatohepatitis: FASCINATE-1, a randomized, placebo-controlled phase 2a trial. Gastroenterology. 2021;161(5):1475–1486.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 17. Syed-Abdul MM 等。脂肪酸合成酶抑制劑 TVB-2640 減少代謝異常男性的肝臟新生脂肪生成。肝臟病學。2020；72(1)：103–118。
       Syed-Abdul MM, et al. Fatty acid synthase inhibitor TVB-2640 reduces hepatic de novo lipogenesis in males with metabolic abnormalities. Hepatology. 2020;72(1):103–118.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 18. Mann JP 等。脂質小滴作為脂肪肝的遺傳核心。肝臟國際期刊。2022;42(12):2594–2596。
       Mann JP, et al. Lipid droplets as the genetic nexus of fatty liver. Liver Int. 2022;42(12):2594–2596.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 19. Romero-Gómez M 等。通過飲食、身體活動和運動治療非酒精性脂肪肝病。肝臟學報。2017;67(4):829–846。
       Romero-Gómez M, et al. Treatment of NAFLD with diet, physical activity and exercise. J Hepatol. 2017;67(4):829–846.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 20. Petersen KF 等。骨骼肌胰島素抵抗在代謝綜合症發病機制中的作用。美國國家科學院院刊。2007;104(31):12587–12594。
       Petersen KF, et al. The role of skeletal muscle insulin resistance in the pathogenesis of the metabolic syndrome. Proc Natl Acad Sci U S A. 2007;104(31):12587–12594.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 21. Viljanen APM 等。體重減輕對肝臟游離脂肪酸攝取和肝臟胰島素抵抗的影響。臨床內分泌與新陳代謝雜誌。2009;94(1):50–55。
       Viljanen APM, et al. Effect of weight loss on liver free fatty acid uptake and hepatic insulin resistance. J Clin Endocrinol Metab. 2009;94(1):50–55.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 22. Balakrishnan M 等。針對非酒精性脂肪肝病患者的行為減重干預：系統性範疇回顧。Hepatol Commun。2023；7(8)：e0224。
       Balakrishnan M, et al. Behavioral weight-loss interventions for patients with NAFLD: a systematic scoping review. Hepatol Commun. 2023;7(8):e0224.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 23. Timlin MT 等。增加的飲食基質供應改變健康男性的肝臟脂肪酸回收。Diabetes。2005；54(9)：2694–2701。
       Timlin MT, et al. Increased dietary substrate delivery alters hepatic fatty acid recycling in healthy men. Diabetes. 2005;54(9):2694–2701.
       通過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 24. Barrows BR, Parks EJ。不同脂肪酸來源對空腹和進食狀態下極低密度脂蛋白-三酸甘油脂的貢獻。J Clin Endocrinol Metab。2006;91(4):1446–1452。
       Barrows BR, Parks EJ. Contributions of different fatty acid sources to very low-density lipoprotein-triacylglycerol in the fasted and fed states. J Clin Endocrinol Metab. 2006;91(4):1446–1452.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 25. Smith GI 等人。胰島素抗性驅動非酒精性脂肪肝病中的肝臟新生脂肪生成。J Clin Invest。2020；130(3)：1453–1460。
       Smith GI, et al. Insulin resistance drives hepatic de novo lipogenesis in nonalcoholic fatty liver disease. J Clin Invest. 2020;130(3):1453–1460.
       通過以下方式查看本文：JCI CrossRef PubMed Google Scholar
       View this article via: JCI CrossRef PubMed Google Scholar

1. 26. Schwarz JM 等人。人類短期碳水化合物能量攝入的變化。對肝臟葡萄糖生成、新生脂肪生成、脂解和全身燃料選擇的顯著影響。J Clin Invest。1995；96(6)：2735–2743。
       Schwarz JM, et al. Short-term alterations in carbohydrate energy intake in humans. Striking effects on hepatic glucose production, de novo lipogenesis, lipolysis, and whole-body fuel selection. J Clin Invest. 1995;96(6):2735–2743.
       通過以下方式查看本文：JCI CrossRef PubMed Google Scholar
       View this article via: JCI CrossRef PubMed Google Scholar

1. 27. Browning JD 等人。短期減重與肝臟三酸甘油脂減少：飲食碳水化合物限制的代謝優勢證據。美國臨床營養學雜誌。2011;93(5):1048–1052。
       Browning JD, et al. Short-term weight loss and hepatic triglyceride reduction: evidence of a metabolic advantage with dietary carbohydrate restriction. Am J Clin Nutr. 2011;93(5):1048–1052.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 28. Hudgins LC 等人。用飲食澱粉替代糖後，人類脂肪酸合成減少。美國臨床營養學雜誌。1998;67(4):631–639。
       Hudgins LC, et al. Human fatty acid synthesis is reduced after the substitution of dietary starch for sugar. Am J Clin Nutr. 1998;67(4):631–639.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 29. Schwarz JM 等人。限制飲食果糖對肥胖兒童肝臟脂肪、新生脂肪生成和胰島素動力學的影響。胃腸病學。2017;153(3):743–752。
       Schwarz JM, et al. Effects of dietary fructose restriction on liver fat, de novo lipogenesis, and insulin kinetics in children with obesity. Gastroenterology. 2017;153(3):743–752.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 30. Miles JM 等人。正常與 2 型糖尿病患者的夜間及餐後游離脂肪酸動力學：胰島素敏感化療法的影響。糖尿病。2003;52(3):675–681。
       Miles JM, et al. Nocturnal and postprandial free fatty acid kinetics in normal and type 2 diabetic subjects: effects of insulin sensitization therapy. Diabetes. 2003;52(3):675–681.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 31. Kim SP 等人。夜間游離脂肪酸在飲食誘導的胰島素抵抗和高胰島素血症的長期發展中獨特地升高。美國生理學雜誌內分泌與代謝。2007;292(6):E1590–E1598。
       Kim SP, et al. Nocturnal free fatty acids are uniquely elevated in the longitudinal development of diet-induced insulin resistance and hyperinsulinemia. Am J Physiol Endocrinol Metab. 2007;292(6):E1590–E1598.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 32. Broussard JL 等人。夜間非酯化脂肪酸的升高是飲食誘導的胰島素抵抗中高胰島素血症補償的早期信號。糖尿病學。2015;58(11):2663–2670。
       Broussard JL, et al. Elevated nocturnal NEFA are an early signal for hyperinsulinaemic compensation during diet-induced insulin resistance in dogs. Diabetologia. 2015;58(11):2663–2670.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 33. Parks EJ 等人。低脂肪高碳水化合物飲食對極低密度脂蛋白三酸甘油脂的組裝、產生和清除的影響。臨床研究雜誌。1999;104(8):1087–1096。
       Parks EJ, et al. Effects of a low-fat, high-carbohydrate diet on VLDL-triglyceride assembly, production, and clearance. J Clin Invest. 1999;104(8):1087–1096.
       通過以下方式查看本文：JCI CrossRef PubMed Google Scholar
       View this article via: JCI CrossRef PubMed Google Scholar

1. 34. Santoro N 等人。肥胖青少年的肝臟新生脂肪生成受到 GCKR 基因中的一個常見變異的調節。臨床內分泌與代謝雜誌。2015;100(8):E1125–E1132。
       Santoro N, et al. Hepatic de novo lipogenesis in obese youth is modulated by a common variant in the GCKR gene. J Clin Endocrinol Metab. 2015;100(8):E1125–E1132.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 35. Wang H 等人。FASN 抑制在肝細胞癌的臨床前模型中的治療效果。肝臟學。2022;76(4):951–966。
       Wang H, et al. Therapeutic efficacy of FASN inhibition in preclinical models of HCC. Hepatology. 2022;76(4):951–966.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 36. Sigala DM 等人。消耗含蔗糖或高果糖玉米糖漿的飲料會增加肝臟脂質並降低成人的胰島素敏感性。臨床內分泌與代謝雜誌。2021;106(11):3248–3264。
       Sigala DM, et al. Consuming sucrose- or HFCS-sweetened beverages increases hepatic lipid and decreases insulin sensitivity in adults. J Clin Endocrinol Metab. 2021;106(11):3248–3264.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 37. Chong MF 等人。果糖對餐後脂血症的急性影響機制。美國臨床營養學雜誌。2007;85(6):1511–1520。
       Chong MF, et al. Mechanisms for the acute effect of fructose on postprandial lipemia. Am J Clin Nutr. 2007;85(6):1511–1520.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 38. Mucinski JM 等人。能量限制和運動後的組織學改善：胰島素抵抗在 MASH 恢復中的作用。肝臟學雜誌。2024;81(5):781–793。
       Mucinski JM, et al. Histological improvements following energy restriction and exercise: the role of insulin resistance in resolution of MASH. J Hepatol. 2024;81(5):781–793.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 39. Vitola BE 等人。減重可降低肝臟脂肪並改善肥胖青少年的肝臟和骨骼肌胰島素敏感性。肥胖（銀泉）。2009;17(9):1744–1748。
       Vitola BE, et al. Weight loss reduces liver fat and improves hepatic and skeletal muscle insulin sensitivity in obese adolescents. Obesity (Silver Spring). 2009;17(9):1744–1748.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 40. Hallsworth K 等人。抗阻運動可在不減重的情況下降低非酒精性脂肪肝病患者的肝臟脂肪及其介質。腸道。2011;60(9):1278–1283。
       Hallsworth K, et al. Resistance exercise reduces liver fat and its mediators in non-alcoholic fatty liver disease independent of weight loss. Gut. 2011;60(9):1278–1283.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 41. Kirk E 等人。飲食中的脂肪和碳水化合物在熱量限制期間對胰島素敏感性有不同的影響。胃腸病學。2009;136(5):1552–1560。
       Kirk E, et al. Dietary fat and carbohydrates differentially alter insulin sensitivity during caloric restriction. Gastroenterology. 2009;136(5):1552–1560.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 42. Magkos F 等人。肥胖患者中葡萄糖、脂肪酸和極低密度脂蛋白動力學及多器官胰島素敏感性的重現性。國際肥胖期刊 (倫敦)。2011;35(9):1233–1240。
       Magkos F, et al. Reproducibility of glucose, fatty acid and VLDL kinetics and multi-organ insulin sensitivity in obese subjects with non-alcoholic fatty liver disease. Int J Obes (Lond). 2011;35(9):1233–1240.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 43. Moustaïd N 等人。胰島素在原代培養的人類脂肪細胞中增加脂肪生成酶活性。營養學期刊。1996;126(4):865–870。
       Moustaïd N, et al. Insulin increases lipogenic enzyme activity in human adipocytes in primary culture. J Nutr. 1996;126(4):865–870.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 44. Wang Y 等人。代謝健康表型比基於體重指數的肥胖表型更能預測非酒精性脂肪肝病人群中的亞臨床動脈粥樣硬化。營養前沿。2023;10:1104859。
       Wang Y, et al. Metabolic health phenotype better predicts subclinical atherosclerosis than body mass index-based obesity phenotype in the non-alcoholic fatty liver disease population. Front Nutr. 2023;10:1104859.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 45. Zhou XD 等人。關於 MAFLD 及心血管疾病風險的國際多學科共識聲明。肝臟國際。2023;17(4):773–791。
       Zhou XD, et al. An international multidisciplinary consensus statement on MAFLD and the risk of CVD. Hepatol Int. 2023;17(4):773–791.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 46. Magkos F 等人。體重減輕後肝臟極低密度脂蛋白-三酸甘油脂分泌的減少與循環中瘦素的變化呈負相關。糖尿病肥胖代謝。2010;12(7):584–590。
       Magkos F, et al. Decrease in hepatic very-low-density lipoprotein-triglyceride secretion after weight loss is inversely associated with changes in circulating leptin. Diabetes Obes Metab. 2010;12(7):584–590.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 47. Mittendorfer B 等人。體重減輕對腹部肥胖女性的極低密度脂蛋白-三酸甘油脂和 apoB-100 動力學的影響。美國生理學雜誌內分泌與代謝。2003;284(3):E549–E556。
       Mittendorfer B, et al. Effect of weight loss on VLDL-triglyceride and apoB-100 kinetics in women with abdominal obesity. Am J Physiol Endocrinol Metab. 2003;284(3):E549–E556.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 48. Browning JD 等人。低碳水化合物飲食迅速且顯著地減少肝內三酸甘油脂含量。肝臟學。2006;44(2):487–488。
       Browning JD, et al. A low-carbohydrate diet rapidly and dramatically reduces intrahepatic triglyceride content. Hepatology. 2006;44(2):487–488.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 49. Ordovas JM。基因對血脂和心血管疾病風險的影響：初級預防的工具。美國臨床營養學雜誌。2009;89(5):1509S–1517S。
       Ordovas JM. Genetic influences on blood lipids and cardiovascular disease risk: tools for primary prevention. Am J Clin Nutr. 2009;89(5):1509S–1517S.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 50. Hu Y 等人。以少數族裔為中心的血脂水平元分析通過基因組學和流行病學（PAGE）研究識別出人口結構中的新位點。PLoS Genet. 2020;16(3):e1008684。
       Hu Y, et al. Minority-centric meta-analyses of blood lipid levels identify novel loci in the population architecture using genomics and epidemiology (PAGE) study. PLoS Genet. 2020;16(3):e1008684.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 51. Larson-Meyer DE 等人。限制熱量攝入與否對超重者的胰島素敏感性、β細胞功能、脂肪細胞大小和異位脂質的影響。糖尿病護理。2006;29(6):1337–1344。
       Larson-Meyer DE, et al. Effect of calorie restriction with or without exercise on insulin sensitivity, beta-cell function, fat cell size, and ectopic lipid in overweight subjects. Diabetes Care. 2006;29(6):1337–1344.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 52. Sato F 等人。飲食誘導的中度減重對肥胖者的肝內和肌肉內三酸甘油脂及葡萄糖代謝的影響。臨床內分泌與代謝雜誌。2007;92(8):3326–3329。
       Sato F, et al. Effects of diet-induced moderate weight reduction on intrahepatic and intramyocellular triglycerides and glucose metabolism in obese subjects. J Clin Endocrinol Metab. 2007;92(8):3326–3329.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 53. Klein S 等人。胃旁路手術改善與非酒精性脂肪肝病相關的代謝和肝臟異常。胃腸病學。2006;130(6):1564–1572。
       Klein S, et al. Gastric bypass surgery improves metabolic and hepatic abnormalities associated with nonalcoholic fatty liver disease. Gastroenterology. 2006;130(6):1564–1572.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 54. Gofton C 等人。瘦型代謝相關脂肪肝病。內分泌學與代謝臨床北美。2023;52(3):431–444。
       Gofton C, et al. Lean metabolic-associated fatty liver disease. Endocrinol Metab Clin North Am. 2023;52(3):431–444.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 55. Sookoian S, Pirola CJ。非酒精性脂肪肝病的遺傳易感性。臨床分子肝臟學。2017;23(1):1–12。
       Sookoian S, Pirola CJ. Genetic predisposition in nonalcoholic fatty liver disease. Clin Mol Hepatol. 2017;23(1):1–12.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 56. McDonough DJ 等人。有氧和肌肉增強體育活動、看電視與非酒精性脂肪肝病：CARDIA 研究。臨床醫學雜誌。2023;12(17):5603。
       McDonough DJ, et al. Aerobic and muscle-strengthening physical activity, television viewing, and nonalcoholic fatty liver disease: the CARDIA study. J Clin Med. 2023;12(17):5603.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 57. Trenell MI。行為與肝臟健康：我們是否正處於一場範式轉變的中間？肝臟學雜誌。2024;81(5):770–771。
       Trenell MI. Behaviour and liver health: are we in the middle of a paradigm shift? J Hepatol. 2024;81(5):770–771.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 58. Beals JW 等人。飲食減重引起的代謝功能改善在肥胖和前糖尿病患者中通過運動得到增強。自然代謝。2023;5(7):1221–1235。
       Beals JW, et al. Dietary weight loss-induced improvements in metabolic function are enhanced by exercise in people with obesity and prediabetes. Nat Metab. 2023;5(7):1221–1235.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 59. Lee JJ 等人。棕櫚油酸在脂肪肝疾病中升高，反映肝臟脂肪生成。美國臨床營養學雜誌。2015;101(1):34–43。
       Lee JJ, et al. Palmitoleic acid is elevated in fatty liver disease and reflects hepatic lipogenesis. Am J Clin Nutr. 2015;101(1):34–43.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 60. Ramos-Roman MA 等人。胰島素激活代謝症候群中的血漿非酯化脂肪酸攝取。動脈硬化、血栓與血管生物學。2012;32(8):1799–1808。
       Ramos-Roman MA, et al. Insulin activation of plasma nonesterified fatty acid uptake in metabolic syndrome. Arterioscler Thromb Vasc Biol. 2012;32(8):1799–1808.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 61. Peter A 等人。肝臟脂質組成和硬脂酰輔酶 A 去飽和酶 1 mRNA 表達可從血漿 VLDL 脂肪酸比率中估算。臨床化學。2009;55(12):2113–2120。
       Peter A, et al. Hepatic lipid composition and stearoyl-coenzyme A desaturase 1 mRNA expression can be estimated from plasma VLDL fatty acid ratios. Clin Chem. 2009;55(12):2113–2120.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 62. Jéquier E 等人。人類能量消耗和燃料利用的評估。營養年鑑。1987;7:187–208。
       Jéquier E, et al. Assessment of energy expenditure and fuel utilization in man. Annu Rev Nutr. 1987;7:187–208.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 63. Schneeman BO 等人。正常脂血症人類中，含有載脂蛋白 B-48 和 B-100 的血漿中三酸甘油脂豐富的脂蛋白對含脂餐的反應之間的關係。美國國家科學院院刊。1993;90(5):2069–2073。
       Schneeman BO, et al. Relationships between the responses of triglyceride-rich lipoproteins in blood plasma containing apolipoproteins B-48 and B-100 to a fat-containing meal in normolipidemic humans. Proc Natl Acad Sci U S A. 1993;90(5):2069–2073.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 64. Fabbrini E 等人。患有非酒精性脂肪肝病的肥胖男性和女性的脂肪組織和肝臟脂質動力學的變化。胃腸病學。2008;134(2):424–431。
       Fabbrini E, et al. Alterations in adipose tissue and hepatic lipid kinetics in obese men and women with nonalcoholic fatty liver disease. Gastroenterology. 2008;134(2):424–431.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 65. Barrows BR 等人。在兩種不同的進食方案下，飲食脂肪酸的溢出及血清非酯化脂肪酸用於合成 VLDL-三酸甘油脂的情況。糖尿病。2005;54(9):2668–2673。
       Barrows BR, et al. Spillover of dietary fatty acids and use of serum nonesterified fatty acids for the synthesis of VLDL-triacylglycerol under two different feeding regimens. Diabetes. 2005;54(9):2668–2673.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

1. 66. Timlin MT, Parks EJ. 健康男性餐後狀態下新生脂肪生成的時間模式。美國臨床營養學雜誌。2005;81(1):35–42。
       Timlin MT, Parks EJ. Temporal pattern of de novo lipogenesis in the postprandial state in healthy men. Am J Clin Nutr. 2005;81(1):35–42.
       透過以下方式查看本文：CrossRef PubMed Google Scholar
       View this article via: CrossRef PubMed Google Scholar

67. Parks EJ, Hellerstein MK。主題回顧系列：以患者為導向的研究。最近在肝臟三酸甘油脂和脂肪酸代謝方面的進展，使用穩定同位素標記技術。J Lipid Res. 2006;47(8):1651–1660。
    Parks EJ, Hellerstein MK. Thematic review series: patient-oriented research. Recent advances in liver triacylglycerol and fatty acid metabolism using stable isotope labeling techniques. J Lipid Res. 2006;47(8):1651–1660.
    透過以下方式查看本文：CrossRef PubMed Google Scholar
    View this article via: CrossRef PubMed Google Scholar