Dry Fasting — The Record

Pathway atlas · node 31 of 41

De novo lipogenesis

What fasting does to this node, by modality, by system and by grade. Disconfirming evidence is reported first.

Founder's key components: FASN, ACLY, ACACA (ACC), SCD1, SREBF1 (SREBP) Founder's description: "Synthesizes fatty acids and complex lipids for membranes and signaling molecules."

DRY status

DRY: none located. Query run: the single Russian plus English Filonov-tradition query listed under node 28. No DRY-modality lipogenesis, tracer or VLDL measurement was located. Dedicated DNL DRY queries and the nine non-English channels could not be run. Recorded NOT RUN.

Disconfirming first

D31.1. THE HEADLINE CAVEAT, and it governs this whole node. Every human DNL number retrieved compares the OVERNIGHT POSTABSORPTIVE state with the FED state inside one normal 24-hour cycle. Not one of them is a fasting modality intervention. None of the nine labels applies to the fasted-versus-fed contrast that generates the claim "fasting suppresses de novo lipogenesis in humans." The claim as usually written is true of the postabsorptive state and UNTESTED for TRE, IF-ADF, IF-5:2, PF, FMD, RAMADAN and DRY. No study was located that measured DNL by tracer under any named fasting modality. This is the largest single gap in slice D and it sits on the node the brief flagged as the one where real tracer data exist. The tracer data exist. They are not fasting-modality data.

D31.2. UNRESOLVED CONFLICT: the healthy fasted-state DNL figure. Three reviews, three different thresholds, two of them citing the SAME primary paper.

Two reviews cite the same primary study and print different thresholds (under 3% versus under 5%). The primary paper could not be opened (two publisher hosts refused), so the discrepancy cannot be settled here. Reported unresolved, all three named. Retrieval debt R31.1.

D31.3. Fasting DNL is NOT uniformly low, and in one population the fed-to-fasted DNL swing was ABSENT. Donnelly KL, Smith CI, Schwarzenberg SJ, Jessurun J, Boldt MD, Parks EJ. "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 to 1351. DOI 10.1172/JCI23621. PMID 15864352. In 9 patients with nonalcoholic fatty liver disease, DNL supplied 22.9 +/- 6.2% of FASTING VLDL-triglyceride and 26.1 +/- 6.7% of liver triglyceride. For total triglyceride-rich lipoproteins, the DNL share was 22.2 +/- 8.1% fasting versus 20.4 +/- 5.2% fed, P = 0.29, that is, NOT different. The roughly ten-fold fed-to-fasted DNL swing that drives the general claim did not occur in this group. Reported in the companion commentary (Tamura and Shimomura 2005): DNL was already elevated in the fasted state in these subjects and showed no further postprandial rise.

D31.4. Donnelly 2005 is internally inconsistent and this file does not paper over it. The normalized table gives fasting VLDL DNL 22.9 +/- 6.2% and diet 14.7 +/- 8.5%. The Results text separately gives VLDL DNL 22.4 +/- 8.2% and diet 13.2 +/- 6.5% (P = 0.01). Both sets are in the same paper. Named, not reconciled. Also: fed-state lipoprotein values are for total triglyceride-rich lipoproteins, not VLDL; VLDL was isolated only in the fasting state, so the fed-versus-fasted VLDL comparison does not exist in this study. Also: the Results describe 5 days of continuous labeling while the Methods say infusions began 4 days before biopsy. The timing is not fully consistent in the source.

D31.5. The methodologist's own verdict cuts against treating DNL suppression as a large effect. Hellerstein MK. "Synthesis of fat in response to alterations in diet: Insights from new stable isotope methodologies." Lipids 1996;31(1):S117 to S125. DOI 10.1007/BF02637062. The abstract states that DNL "appears not to be a quantitatively major pathway even under conditions of surplus carbohydrate energy intake." No numerical values were retrievable from the page opened; the full text is paywalled. Retrieval debt R31.2. Recorded because it is the inventor of the MIDA method constraining the magnitude claim.

D31.6. Surrogate markers for fasting DNL are unreliable. Located, not opened. "Fasting hepatic de novo lipogenesis is not reliably assessed using circulating fatty acid markers." Title located; source not opened (host blocked). Retrieval debt R31.3. Recorded because it bears directly on any attempt to infer this node's fasting behavior from fatty acid ratios instead of tracers.

D31.7. The only two located human fasting measurements of a named component of this node are both NULLS, and both are in muscle, not liver.

FASN, ACLY, SCD1 and SREBF1 have NO located human fasting measurement at all.

Graded rows

31-R1. In nonalcoholic fatty liver disease, fasting-state source apportionment of liver and VLDL triglyceride, by triple tracer. Grade B+, tier H3, modality: none of the nine, postabsorptive versus fed. Donnelly KL et al. J Clin Invest 2005;115(5):1343 to 1351. DOI 10.1172/JCI23621. PMID 15864352. 9 NAFLD patients (5 male, 4 female), age 44 +/- 10 years. Tracers: [13C4]potassium palmitate 2.5 micrograms/kg/min from day 1 at 17:00 for the NEFA source; sodium [13C1]acetate 10 g/day from day 1 at 17:00 with mass isotopomer distribution analysis for DNL; [2H31]glyceryl-tripalmitin 1 to 2 g/day with each meal for the dietary source. Fasting blood at 06:45 on days 2 to 5; fed blood at 15:00 (3 h after lunch) on days 2 to 4; liver biopsy day 5 at 08:00. Fast length is NOT stated.

Serum NEFA, percent of triglyceride fatty acids, mean +/- SD:

SourceFastingFedP
Adipose81.7 +/- 8.261.7 +/- 11.90.0003
Diet10.4 +/- 4.926.2 +/- 8.90.0006
DNL7.0 +/- 4.09.4 +/- 4.50.01

Total triglyceride-rich lipoprotein triglyceride:

SourceFastingFedP
Adipose60.4 +/- 18.227.9 +/- 10.20.0001
Diet12.1 +/- 4.119.1 +/- 8.30.01
DNL22.2 +/- 8.120.4 +/- 5.20.29

Fasting-state VLDL triglyceride and liver triglyceride, normalized to 100% of accounted-for fatty acids:

SourceVLDL TG, day 5 fastingLiver TG
NEFA62.4 +/- 11.759.0 +/- 9.9
DNL22.9 +/- 6.226.1 +/- 6.7
Diet14.7 +/- 8.514.9 +/- 7.0

No significant difference between VLDL and liver for any source. Other values: dietary fatty acids entering VLDL via chylomicron remnant uptake 7.2 +/- 5.6% (fasting). Accrual rates: DNL 3.5% per day versus diet 1.7% per day, P = 0.004. Fraction of liver TG-palmitate accounted for after labeling 20.4% to 63.7%. Liver TG fractional synthesis rate 5.1% to 15.5% per day; estimated turnover time 38.4 +/- 15.6 days. Liver TG content undetectable to 0.49 g/g tissue (healthy reference about 0.01 g/g). Fasting serum TG 3.2 +/- 2.2 mmol/L. Fasting VLDL concentration (Sf 60 to 400) 1.0 +/- 0.9 mmol/L. NEFA rate of appearance on day 2, 6.3 +/- 2.9 micromol/kg/min. HOMA-IR 4.1 +/- 2.8 (range 1.2 to 8.9). See D31.4 for this paper's internal inconsistencies.

31-R2. The fasted-state DNL figure in healthy people depends on the diet that PRECEDED the fast, and the dependence is large. Grade B+, tier H3, modality: none of the nine, postabsorptive versus fed under different preceding diets. Schwarz JM, Clearfield M, Mulligan K. J Osteopath Med 2017;117(8):520 to 527. DOI 10.7556/jaoa.2017.102.

ConditionValue reportedPrimary study cited
Healthy, low or normal carbohydrate dietFasting DNL under 3% of VLDL-TG; postprandial 20% to 30%, roughly a 10-fold riseTimlin MT, Parks EJ, Am J Clin Nutr 2005
5 days energy-balanced high-carbohydrate feeding (65% to 70% of calories)Fasting DNL described as quantitatively significant; no numeric value given in the figure captionSchwarz JM et al., Am J Clin Nutr 2003
5 days carbohydrate or fructose overfeedingFractional DNL rises to 20% in the FASTING state and 30% to 35% fedFigure cites Schwarz JM et al., J Clin Invest 1995
Fructose versus glucose, 6 lean adults, 6 h oral feedingFructose: DNL up about 20-fold, about 25% of circulating VLDL-TG from DNL. Glucose: DNL unchanged, 1% to 2% of VLDL-TG de novoHellerstein MK, Schwarz JM, Neese RA, Annu Rev Nutr 1996

THE LOAD-BEARING POINT: the SAME fasting state yields under 3% on a normal-carbohydrate diet and about 20% after 5 days of carbohydrate overfeeding. "Fasting suppresses DNL" is therefore conditional on what the person ate before the fast, by roughly a factor of seven in these data. A surface that states the suppression without that condition is overclaiming. The article gives no absolute DNL rates in g/day. All four rows are second-hand through this review; none of the four primary papers was opened. Retrieval debt R31.4.

31-R3. Fructose loading raises the FASTING DNL figure, measured longitudinally. Grade B, tier H3, modality: none of the nine, dietary intervention. Moore JB, Gunn PJ, Fielding BA. Nutrients 2014;6(12):5679 to 5703. DOI 10.3390/nu6125679.

The review explicitly separates out figures that are NOT fasting data, and this file preserves that separation: the Donnelly 26% figure followed a 4-day acetate infusion and is not described as fasting; a Donnelly follow-up (10% versus 23%; 2.6 versus 1.5 micromol/min) was measured over 12 h after an evening meal; the Lambert et al. control value of 14% palmitate DNL is not specified as fasting.

31-R4. Reference record for the primary fasted-versus-fed DNL study, which was NOT opened. Timlin MT, Parks EJ. "Temporal pattern of de novo lipogenesis in the postprandial state in healthy men." Am J Clin Nutr 2005;81:35 to 42. The publisher's article record (journal, year, volume, pages) was retrieved through a citation lookup and is witness-level for the RECORD under the brief's standard. The article itself was not opened: two publisher hosts refused. Every number attributed to it in this file is second-hand through three reviews that disagree with each other (see D31.2). Retrieval debt R31.1.

Open questions, node 31


← 30. Serine–glycine / one-carbon (folate) metabolism All 41 nodes 32. TCA-cycle enzymes and 2-oxoglutarate signaling →


dryfasting.com has been published since 2014. This evidence record was compiled on 22 November 2026. Every row names the fasting modality the study actually ran, the system it was measured in and the grade. Where no source carries a DOI or PMID the row says so rather than supplying one. Nothing is sold on this site.