Patients with HTG-AP have high serum TGs, serum NEFAs, and worse clinical outcomes. Between August 2019 and November 2021, 488 patients presented to the emergency room (ER) of Mayo Clinic Arizona with lipase levels of more than 3 times the upper limit of normal (ULN). Of these, 201 patients were excluded, as they did not fulfill the diagnostic criteria of AP. A flowchart showing the selection of the study population is shown in Figure 1B. Of the 287 patients with AP, 18 individuals were excluded because there were no admission TG measurements or their serum samples were duplicates. Based on chart review, 27 patients had HTG-AP and 242 patients had non–HTG-AP. As shown in Table 1, age, sex distribution, BMI, diabetes status (data not shown), race, median of abdominal pain duration, and time from sample collection to processing were similar between the patients with HTG and the non–HTG-AP patients. The HTG-AP group had fewer individuals with alcoholic AP (3) and biliary AP (1) than did the non–HTG-AP group (54 and 48, respectively; P < 0.001). Eight patients with HTG-AP had documented hyperlipidemia before admission for AP. Six of these were on a statin, among whom 4 were also on fenofibrate, and 1 was on gemfibrozil. Only 1 patient with HTG-AP was suspected to have a genetic cause, but this was not confirmed.

Table comparing HTG-AP versus non–HTG-AP patient populations

We first compared serum TGs and NEFAs among the patients. As shown in Table 1, patients with HTG-AP had significantly elevated median serum TG levels [730 mg/dL (IQR: 561–1,064) or 8.5 mM (6.6–12.5) versus 161 mg/dL (IQR: 111–238) or 1.9 mM (1.3–2.8) in non–HTG-AP patients; P < 0.0001]. Patients with HTG-AP had significantly higher admission serum NEFAs of 0.95 mM (0.6–1.5 mM) compared with non–HTG-AP patients, who had median NEFAs of 0.61 mM (0.37–0.87 mM; P < 0.0001). All the principal long-chain NEFAs were increased in the patients with HTG-AP. These included palmitic acid, its unsaturated product palmitoleic acid, and also stearic acid, oleic acid, and linoelic acid.

On comparing clinical outcomes, patients with HTG-AP had significantly higher organ failure rates (30% vs. 10%; P:0.02) and of severe AP (19% vs. 7%; P:0.03) compared with non–HTG-AP patients. We observed no difference in early severity (i.e., within the first week) (30) between the 2 groups (3 of 5 in HTG-AP vs. 8 of 16 in the non–HTG-AP group; P = 1.0). Fifteen of 153 (9.8%) males, and 6 of 116 females (5.1%) developed severe AP (P = 0.18). Patients with HTG-AP also had higher intensive care unit (ICU) admission rates (26% vs. 8%; P:0.003) and median length of stay (5 vs. 3 days; P:0.0002) compared with non–HTG-AP patients. Two of the 27 patients with HTG-AP had pancreatic necrosis compared with 13 of 242 non–HTG AP patients (P = 0.63).

We thus went on to look for evidence supporting lipolytic generation of NEFAs during HTG-AP.

Lipolysis of circulating TGs by pancreatic lipase generates injurious unsaturated NEFAs. To determine the role of lipolytic NEFA generation on HTG-AP severity, we studied whether organ failure, which defines pancreatitis severity in humans (30), could be induced in experimental models by intravascular lipolysis during hypertriglyceridemia. For this we first induced HTG (Figure 2, A–G) in the UFA-fed C57bl/6 mice (WT) by administering poloxamer-407 to these mice on day –1 (as described in Methods). We then induced pancreatitis (HTG + AP) using IL-12 and IL-18 in some of these (red symbols in Figure 2, A–G), whereas others were followed with only HTG (blue symbols). We also induced similar HTG-AP mice in UFA-fed mice that had genetic deletion of the pancreatic TG lipase (PNLIP) gene, herein referred to as PTL-KO mice (green symbols in Figure 2, A–G). Mice with HTG alone (blue symbols in Figure 2, A–G) were used as controls. Mice in all groups had increased serum TG levels from less than 100 mg/dL at baseline to the 15,000–25,000 mg/dL range within 1 day of HTG induction. Serum TG levels were similar in all groups before AP induction (day 0 in Figure 2B). These were 18,667 ± 2,402 mg/dL in HTG-only controls, 21,588 ± 4,245 mg/dL in C57bl/6 mice before HTG-AP, and 19,597 ± 4,450 om PTL-KO mice (P = 0.83). Mice with AP (induced on day 0 after blood collection) had a significant increase in serum lipase by day 1 (Figure 2, A and B). AP caused a rapid reduction in serum TGs (red dots in pink oval, Figure 2B) but not in mice with HTG alone (blue, Figure 2B) or in PTL-KO mice (green dots, Figure 2B). The reduction in TGs resulted in a corresponding significant increase in serum NEFAs (red dots in pink oval, Figure 2C) on day 1 of AP. Mice with HTG alone had no change in serum TGs or NEFAs on day 1 (blue dots in blue ovals in Figure 2, B and C). Similarly, PTL-KO mice had no change in serum NEFAs on day 1 of AP (green dots Figure 2C). Between day 1 and 2, WT mice with AP began to appear moribund and required euthanasia. WT mice with HTG-AP had higher serum creatinine levels at euthanasia (Figure 3D) and a preterminal reduction in pulse distention consistent with hypotension (19, 31) (Figure 3E). WT mice also developed generalized hypothermia (Figure 3F), along with an 80% reduction in survival by day 3 (Figure 3G). These findings are consistent with HTG-AP induction of multisystem organ failure. WT mice had 4.7% ± 5.3% pancreatic parenchymal necrosis, seen as pale pink areas with loss of cellular detail (arrows in Figure 2H). This necrosis was reduced to 0.6% ± 0.7% (P < 0.01) in PTL-KO mice (Figure 2, I and J). The pancreatic edema score averaged 2/4 in both groups of mice (Figure 2K).

Parameters of mice with HTG alone or HTG-AP, those given NEFAs, and in vitro studies using pancreatic acini. (A–H) Mice were given poloxamer-407 on day –1 to induce HTG (H) alone (blue). HTG-AP (HAP) was induced on day 0 (i.e., 1 day after poloxamer-407) in C57bl/6 WT mice (red) or pancreatic TG lipase–KO mice (PTL KO) (green dots). Dot plots with means are shown for serum lipase (A), serum TGs (B), and serum NEFAs (C) for the days indicated on the x axis. Serum creatinine at necropsy (D), along with carotid pulse distention (E) and rectal temperature (F) recorded before euthanasia are shown. (G) Comparison of the survival curves for the 2 groups with P values based on the log-rank (Mantel-Cox) test. (H and I) Images of H&E-stained pancreatic sections from WT mice (H) and PTL-KO mice (I) with HAP. Scale bars: 200 μm. (J and K) Dot plots showing pancreatic necrosis (J) and pancreatic edema (K) in WT and PTL-KO mice with HAP. (L–N) Dot plots with means comparing the effects of administering linoleic acid (red dots) or palmitic acid (blue dots) on serum BUN levels at necropsy (L), along with carotid pulse distention (M) and rectal temperature (Temp) (N) recorded before euthanasia. P values were determined by 2-way ANOVA was done (A–F), log-rank (Mantel-Cox) test (G), Mann-Whitney U test (J), 2-tailed Student’s t test was done (K), and ordinary 1-way ANOVA (L–N).

Effect on rats of intravenous infusion of TGs versus infusion of TGs with PPL, with or without the lipase inhibitor orlistat. Bar graphs (with SD and individual values) comparing the effects of infusion of PPL; TG (GTO); PPL plus TG; and PPL plus TG, plus the lipase inhibitor orlistat at baseline (Bas) and after infusion. Parameters are serum lipase (A), NEFAs (B), BUN (C), and ionized calcium (iCa) (D). (E) Representative images of H&E-stained images of pancreatic sections from rats infused with GTO alone or GTO plus PPL. Scale bars: 200 μm. (F–H) Plots of preterminal oxygen saturation (Ox sat) (F), LDH activity in BAL fluid from the lungs (G), and protein concentrations in the BAL (H). (I) Representative lung histologic images after H&E staining from each group mentioned at the top of the image. Scale bars: 100 μm. The rectangle inset is zoomed outside the ×20 image. In the PPL plus GTO group, arrows point to alveolar wall damage, and asterisks show fluid-filled alveoli. All data in graphs were compared by ordinary 1-way ANOVA with multiple comparisons.

To determine the specific roles of unsaturated NEFAs, we compared organ failure parameters in mice given the saturated NEFA palmitic acid versus the unsaturated NEFA linoleic acid. Both of these NEFAs were increased in patients with HTG-AP (Table 1). Only linoleic acid increased blood urea nitrogen (BUN) levels and lowered carotid pulse distention, which indicated hypotension (19, 31), worsened hypothermia (Figure 2, L–N), and increased apoptotic cells in the lungs, consistent with acute lung injury (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI184785DS1), resulting in a moribund appearance of the mice, requiring euthanasia by day 3. Therefore, lipolytic release of unsaturated NEFAs like linoleic acid may worsen AP outcomes.

Severe AP includes respiratory and renal failure. We thus studied whether intravascular lipolysis of TGs worsens HTG-AP in a distinct and direct intravascular lipolysis model of HTG in rats. For this, we intravenously infused TGs (glyceryl trioleate [GTO], TGs composed of 3 oleic acid chains) into precannulated rats, alone or with PPL and the lipase inhibitor orlistat. GTO infusion increased serum TGs from 94 ± 29 mg/dL to 233 ± 120 mg/dL (P = 0.0004; Supplemental Figure 2A). As shown in Figure 3, PPL infusion increased lipase (Figure 3A), which was inhibited by orlistat. As shown in Figure 3B, the PPL + GTO group had the highest NEFA levels, consistent with lipolysis of TGs, which was prevented in the PPL + GTO + orlistat group. The PPL + GTO group also had significantly higher serum BUN levels (Figure 3C) and lower ionized calcium levels (Figure 3D), which were prevented by orlistat. PPL + GTO treatment only induced edema in the pancreas but no necrosis or inflammation, whereas levels in other groups remained similar to those in controls (Figure 3E and Supplemental Figure 2B). While infusion of PPL or GTO did not cause any changes in oxygen saturation, infusion of PPL + GTO together caused a significant decrease in the percentage of oxygen saturation (Figure 3F) compared with baseline, requiring euthanasia. This was associated with an increase in lactate dehydrogenase (LDH) levels and protein concentrations in the bronchoalveolar lavage (BAL) fluid, supporting lung injury (Figure 3, G and H). Histologically the PPL + GTO group had fluid in the alveoli that was seen as diffuse pink staining (asterisks in Figure 3I), along with damage to the alveolar walls (arrows in Figure 3I, PPL + GTO panel). However, infusing orlistat in addition of PPL and GTO (PPL + GTO + orlistat) prevented the drop in oxygen saturation, the LDH and protein increase, and histological changes noted in the PPL + GTO group. Thus, intravascular lipolysis of circulating TGs can cause lung injury and organ failure with minimal pancreatic injury during severe AP.

The correlation of serum lipase with NEFAs increases by factoring in the corresponding TGFAs. We next aimed to understand the relation between circulating NEFAs and TGs, i.e., whether TG lipolysis leads to NEFA formation in AP or if the FAs in TGs (i.e., TGFAs) merely correlate with their precursor NEFAs. For this, we first correlated specific NEFAs to their TGFA concentrations for all patients, which would support synthesis via the Kennedy pathway (Figure 4A). To determine whether TGFA hydrolysis by lipase generates NEFAs, as would happen during AP, we factored in serum lipase levels while correlating individual NEFAs to the corresponding TGFAs, i.e., we correlated NEFAs to TGFA × lipase. We also correlated NEFA concentrations to the corresponding serum lipase activity for all patients.

Effect of lipase activity on the relationship between individual NEFAs and TGFAs in patients with pancreatitis. (A) Schematic comparing the Kennedy pathway (left side, green background) by which NEFAs are physiologically incorporated into the TGs for storage (adipose, liver) or transport (intestine) versus the pathological release of NEFAs from intravascular TG lipolysis by pancreatic lipases during HTG-AP (red background on the right side). (B) Correlation of serum lipase activity with individual TGFAs for all patients. Each column shows a unique FA. The upper value shows the correlation coefficient, and the lower number the P value. (C) All patient data were formatted as in B. Middle row (dark gray background) correlates individual serum NEFAs with serum lipase; top row (white background) correlates individual NEFAs and their TGFA concentrations; and bottom row (white background) correlates individual NEFAs and the product of serum lipase × TGFA concentrations. The light gray rows show P values comparing the strength of correlations (COCOR as described in Methods) between the middle row and corresponding top or bottom rows. Those with a P value of less than 0.05 are shown in red. (D) Bar graphs of correlations (R values) arranged by serum TG concentrations (x axis) for individual FAs. Each graph is for a FA (mentioned above) and shows correlations of its NEFAs with corresponding TGFAs (back bars) or NEFAs with the product of the corresponding TGFA concentration × serum lipase (red bars). Asterisks show the bar with significantly stronger correlations versus normal TGs, i.e., TGs below 150 mg/dL. All correlations are Spearman correlations, and P values are 2 tailed. The comparison of COCOR correlations between 2 Spearman coefficients were done as described in Methods, and P values are shown. Asterisks in D indicate a COCOR P value of less than 0.05 versus the normal (<150 mg/dL) TG group.

On initial analysis, abnormal distributions were noted for individual NEFAs (Supplemental Figure 3), and also for serum lipase (i.e., enzyme), TGFAs (the substrate), and the multiplicative product of (TGFA × lipase) (Supplemental Figure 4). While serum lipase did not correlate (Spearman) with any TGFA concentration (Figure 4B), serum lipase in AP did have a weak but significant Spearman correlation with each of the major NEFAs measured (dark gray, middle row, Figure 4C). We then correlated the concentrations of specific NEFAs to their corresponding TGFAs for all patients with AP. NEFA concentrations of palmitic acid (PA) (C16:0) and palmitoleic acid (POA) (C16:1), correlated more strongly with corresponding TGFAs (upper row, Figure 4C) than with serum lipase. In this NEFA-versus-TGFA correlation for PA, POA was not improved by multiplying the TGFA concentrations by lipase (bottom row, Figure 4C), despite the latter remaining stronger than the correlation with lipase. Therefore, PA and POA in the NEFA and TGFA fractions correlated independently of lipolysis, consistent with the Kennedy pathway contributing to their relationship in all patients with AP. However, for total NEFAs and, specifically, for linoleic acid (C18:2) NEFAs, the correlation improved significantly after multiplying their TGFA concentrations by serum lipase (P = 0.03, bottom row in Figure 4C).

To determine the relevance of these correlations to HTG-AP, we stratified the correlations by serum TG concentrations in mg/dL (Figure 4D). For this, serum TG levels below 150 (n = 107; or controls), 151–300 (n = 101), or 301–500 (n = 35), and above 500 mg/dL (n = 27) were arranged on the x axis with Spearman correlations on the y axis. The mean, median, and other descriptors of these groups are shown in Supplemental Figure 5. The black bars in Figure 4D depict NEFA-to-TGFA correlations (relevant to the Kennedy pathway). The red bars show NEFA correlations to lipase × TGFAs (supporting lipolysis). While NEFA-to-TGFA correlations of unsaturated palmitoleate, oleate, and linoleate increased up to TGs below 500 mg/dL (black asterisk in Figure 4D), these correlations weakened at TG values above 500 mg/dL. Interestingly, at TG levels above 500 mg/dL, factoring in the lipase activity (red bars) further strengthened the NEFA-to-lipase × TGFA correlations of POA (r = 0.8), linoleic acid (LA) (r = 0.62), and oleic acid (OA) (r = 0.52) versus corresponding correlations at TGs below 150 mg/dL (all with P < 0.05, red asterisk). Therefore, during AP, lipolysis of linoleate, palmitoleate, and oleate from their TGs increased with lipase activity at TGs of greater than 500 mg/dL. The weaker TGFA-NEFA correlations at TGs above 500 mg/dL support this. Interestingly, neither correlation increased with TG concentrations for the saturated FAs stearic acid (SA) and PA. These findings supporting lipolytic generation of unsaturated NEFAs from TGs during AP and the established injurious roles of lipolytically generated unsaturated NEFAs (21, 24, 32) are consistent with HTG-AP being more severe (2, 7, 8, 14, 33–35).

We also correlated the proportions of individual NEFAs versus TGFAs in relation to serum TG levels, double-bond number (i.e., unsaturation), and FA chain length (Supplemental Figure 6). When comparing the correlations (COCOR) among FAs based on the double-bond number using R (α 0.05; CI, 0.95) (36, 37), we noted Pearson coefficients to be significantly higher for linoleic acid than oleic acid or stearic acid (which have 18 carbon atoms; P < 0.001). Among 16 carbon FAs, COCOR for POA was stronger than palmitic acid (P < 0.001; Supplemental Figure 6, A–F). These findings are consistent with those of previous studies showing that unsaturation improves the fit of TGs in pancreatic lipases (23). All FAs except stearic acid had stronger correlations at TGs above 200 mg/dL (n = 111 patients, Supplemental Figure 6, G–L). Equivalent results were obtained when the cutoff was greater than 150 mg/dL (n = 162, data not shown). Overall, these findings agreed with breakdown of TGFAs contributing to the unsaturated NEFAs increased during HTG-AP.

Last, to experimentally verify the clinical observations, we used TGs containing linoelic acid, oleic acid, palmitic acid, and palmitoleic acid in various combinations and exposed them to lipases released by acinar cells in suspension. As shown in Supplemental Figure 7A, lipolysis of the TGs of palmitic acid (C16:0), i.e., tripalmitin (PPP) was negligible and increased significantly by replacing 2 acyl chains in the TGs with POA (C16:1; PO) in PO(2)-P. Similarly, addition of the saturated palmitate (C16:0) to the TGs of linoleate or oleate (i.e., using 1,2-dilinoleoyl-3-palmitoyl-rac-glycerol [LLP] or 1,2-dioleoyl-3-palmitoyl-rac-glycerol [OOP] in place of glyceryl trilinolein (LLL) or triolein (OOO) also reduced lipolysis of their TGs by 50%–80% (Supplemental Figure 7B). Moreover, replacement of 1 linoleic acid in LLP with an oleic acid, i.e., LOP, further reduced TG lipolysis by more than 90%. Therefore, while linoleic acid (C18:2) had the highest fidelity of lipolysis by pancreatic lipases, adding oleic acid (C18:1) and palmitic acid (C16:0) incrementally decreased this fidelity. These results support the concept that double bonds (i.e., unsaturation) in long-chain FAs increased their lipolysis from a TG and that saturation interfered with TG lipolysis (23). Overall, our results show that excessive generation of unsaturated NEFAs from circulating TG lipolysis may have caused organ failure and worsened HTG-AP severity.