Glucagon-like peptide-1 and glucagon-like peptide-2 regulation during human liver regeneration – Scientific Reports

Patient demographics

To study perioperative GLP-1 and GLP-2 plasma concentration dynamics, we included 46 patients, undergoing minor (n = 26) or major (n = 20) hepatic resections. Twenty patients had hepatic resection for metastasized colorectal carcinoma (mCRC), 10 for hepatocellular carcinoma (HCC), 12 for intrahepatic or perihilar cholangiocarcinoma carcinoma (iCCA/pCCA), 1 for neuroendocrine liver metastasis (NELM), 1 for non-neuroendocrine non-colorectal (NNECR) liver metastasis and 2 patients had benign liver lesions. Seven patients developed PHLF, of whom 3 died from liver failure. Patient demographics are shown in Table 1.

GLP-2 plasma levels increase postoperatively, but not conditionally upon resection extent

First, we wanted to evaluate if GLP-1 and GLP-2 plasma concentrations in patients are altered upon liver resection (LR). Additionally, we aimed to explore possible perioperative dynamics and if differences are related to the extent of lost liver tissue. Therefore, we excluded patients who developed PHLF and/or postoperative morbidity (Dindo ≥ 3) and explored the influence of the resection extent (major LR vs. minor LR) on GLP-1 and GLP-2 plasma concentration dynamics and compared plasma levels at perioperative timepoints.

GLP-1 plasma levels did not change according to baseline concentrations following liver surgery, regardless of the resection extent. Although in minor LR GLP-1 level decreased from POD1 to POD5, but between minor and major LR no differences could be observed on any perioperative timepoints (Fig. 1a–d, Supplementary Materials & Methods Table S1). Remarkably, GLP-2 levels increased 1.9-fold in both, minor and major LR, within the first postoperative day. In minor LR GLP-2 reached baseline level until POD5, whereas in major LR GLP-2 level elevation remained (Fig. 1e–h). Taken these findings together, an early postoperative, physiological increase of GLP-2 plasma concentrations in all patients can be observed, regardless of the resected liver volume, indicating an upregulation during liver regeneration, timely persisting in resections of greater extent.

PHLF is associated with decreased GLP-1 and increased GLP-2 plasma concentrations

Next, we focused on perioperative GLP-1 and GLP-2 plasma concentration dynamics in patients who developed PHLF. We observed, that patients who developed PHLF displayed 2.3-fold higher GLP-1 plasma concentrations preoperatively, compared to patients without PHLF (p = 0.030). Further a trend towards a postoperative decline until POD1 was present in both, but 3.2-times more pronounced in patients who developed PHLF (p = 0.027) (Fig. 2a–c, Supplementary Materials & Methods Table S2). Until POD5, GLP-1 level of patients with PHLF seem to recover to baseline levels (Fig. 2c,d). Although, given the pronounced early postoperative GLP-1 level alterations in PHLF, the AUC of GLP-1 plasma level trajectories, were similar (p = 0.114) (Fig. 2c, Supplementary Materials & Methods Table S3).

Perioperative GLP-1 and GLP-2 concentrations in the total cohort regarding PHLF and in postoperative morbidity when PHLF was excluded; Differences in GLP-1 and GLP-2 plasma levels between groups and perioperative timepoints are indicated (a, e, i, m). Individual perioperative dynamics are shown (b, f, j, n). Graphs with error-bars illustrate perioperative GLP level dynamics, columns with error- bars exemplify differences in the AUCs (c, g, k, o). Postoperative changes of GLP-1 and GLP-2 plasma level until POD5 are depicted as fold-change compared preoperative values (d, h, l, p); Man-Whitney-U test; Wilcoxon signed-rank test; *p < 0.05; **p < 0.005; ***p < 0.0005.

In contrast, preoperative GLP-2 plasma concentrations did not differ, but increased 3.4-fold within the first postoperative day in patients with PHLF, compared to an 1.5-times increase in patients who did not develop PHLF, resulting in 2.0-times higher GLP-2 level in PHLF on POD1 (p = 0.008). This difference became even more apparent, as in patients without PHLF, GLP-2 levels decreased until POD5 but remained 2.6-fold higher in patients who developed PHLF (p = 0.006) (Fig. 2e–h, Supplementary Materials & Methods Table S2). This dynamic could be further illustrated in a significantly greater AUC (p < 0.001) and GLP-2 level ratios at specific timepoints compared to baseline in patients who developed PHLF (Fig. 2g + h, Supplementary Materials & Methods Table S3).

These results not only show a disconnect between GLP-1 and GLP-2 dynamics in the early postoperative period, as further illustrated by the absence of correlations at perioperative time points (Supplementary Materials & Methods Fig. S1), but also indicate a persistent up-regulation of GLP-2 in the context of PHLF.

Perioperative GLP-1 and GLP-2 dynamics are not associated with postoperative morbidity

Assuming PHLF to be a discrete entity, which secondary leads to morbidity and mortality, we aimed to evaluate, if differences in perioperative GLP concentration dynamics are caused by postoperative complications independently of PHLF. Therefore, we excluded all PHLF patients and compared GLP-1 and GLP-2 plasma levels between patients with morbidity ≥ 3 and patients with an uneventful postoperative course.

In the postoperative course, we did not observe significant dynamics of GLP-1 concentrations in patients with severe morbidity, neither were GLP-1 plasma concentration differences present on respective timepoints (Fig. 2i–l). On the other hand, postoperative GLP-2 concentrations increased in both groups, remaining elevated in patients with severe postoperative morbidity, while in patients with an uneventful postoperative course, the levels further decreased to baseline. No significant differences were observed in the trajectories of GLP-2 plasma concentrations or at respective time points (Fig. 2m–p, Supplementary Materials & Methods Table S4). Summarizing these results, it is conceivable, that postoperative GLP-1 and GLP-2 dynamics are not solely a mechanism in the context of severe postoperative complications but a feature of a dysfunctional regeneration of the liver itself.

GLP-1 but not GLP-2 baseline plasma level are associated with obesity, and histological features of MASLD

Given to previous reports that GLP plasma concentrations are altered in obesity and MASLD, which are frequently associated with diabetes, we therefore thought to investigate whether baseline GLP-1 or GLP-2 plasma level are associated with these conditions. Patients with BMC ≥ 25 kg/m² presented higher GLP-1 (p = 0.041), but not GLP-2 levels (Supplementary Materials & Methods Fig. S2b). Similar alterations were observed in histopathological findings of steatosis (p = 0.004) and inflammatory activity (p = 0.040) (Supplementary Materials & Methods Fig. S2c + d). The presence of fibrosis of any stage was not associated with differences in GLP-1 and GLP-2 levels (Supplementary Materials & Methods Fig. S2e). Patients with MASLD on the other hand, rendered a non-significant trend towards elevated preoperative GLP-1 and GLP-2 levels. Importantly, no associations of GLP-1 and GLP-2 concentrations at baseline were seen regarding underlying diabetes mellitus (Supplementary Materials & Methods Fig. S2a). These findings indicate, that patients who suffer from obesity or an underlying liver disease, displayed higher preoperative GLP-1 plasma levels, whereas GLP-2 levels do not appear to be affected.

Postoperative GLP-1/GLP-2 concentrations appear production controlled and not degradation associated

Given previous reports of bile acids and IL-6 having profound effects on GLP secretion of enteroendocrine L-cells, our next focus was on dynamics of plasma bile acids and IL-6 concentrations, particularly their perioperative dynamics.

We observed no difference in plasma bile acid concentrations between patients with and without PHLF. Neither a difference at indicated timepoints, nor dynamics between timepoints or patients were observed (Fig. 3a–c). Furthermore, when examining the influence of the extent of resection on changes in bile acid concentrations, after excluding patients who developed PHLF or experienced morbidity of ≥ 3, we did not observe any differences in bile acid concentrations or their postoperative dynamics (Fig. 3c).

Perioperative bile acids, IL-6 and DPP4 plasma levels. Comparison of bile acids (a), IL-6 (d) and DPP4 (g) plasma concentrations regarding PHLF on and between perioperative timepoints. Violine blots with horizontal and dotted lines for means and quartiles. Dynamics illustrated in graphs with error- bars and comparison of AUCs displayed by columns with error bars (b, e, h). Postoperative differences of bile acid (c), IL-6 (f) and DPP4 (i) concentrations associated with the resection extent solely (major vs. minor resections, PHLF excluded), are displayed between and on specific timepoints. Mann–Whitney-U test; Wilcoxon signed-rank test; IL-6 Interleukin-6, DPP4 dipeptidyl peptidase 4, *p < 0.05; **p < 0.005; ***p < 0.0005.

IL-6 levels exhibited an early postoperative increase in both groups, with no differences regarding PHLF development (Fig. 3d). In the further postoperative time course, IL-6 level remained elevated in PHLF whereas in patients without PHLF, IL-6 decreased, resulting in a pronounced elevation of IL-6 in PHLF on POD5 (p < 0.001). These dynamics were also mirrored by a significant difference in the AUC of perioperative IL-6 trajectories (p = 0.005) (Fig. 3d + e). Once again, after excluding patients with PHLF or morbidity ≥ 3, no disparities in postoperative IL-6 levels associated with the extent of resected liver tissue were apparent (Fig. 3f). These results indicated, that impaired liver regeneration rather than the resection extent is associated with elevated postoperative IL-6 levels.

DPP4 is highly expressed in the liver, but, as our observation of immunohistochemical DPP4 staining revealed, mainly located on the biliary compartment of hepatocytes and not in the sinusoids (Supplementary Materials & Methods and Fig. S3). Given the fact that the soluble enzymatically active form of DPP4 might be responsible for the extensive degradation of GLP-1 and GLP-2 in the portal circulation, we sought to evaluate if changes in circulating DPP4 levels are associated with distinct GLP-1 or GLP-2 dynamics in PHLF. We did not observe any differences in DPP4 levels at any timepoints, nor did we notice any remarkable dynamics within the first 5 postoperative days in patients with PHLF (Fig. 3g,h). After excluding patients with PHLF and morbidity ≥ 3, we compared DPP4 dynamics according to extent of resected liver volume. We could observe an initial postoperative decrease of circulating DPP4 level in both, minor and major resections, but no differences at certain timepoints (Fig. 3i). Summarized, these findings point out, that DPP4 levels are somehow affected during the postoperative course, but hepatic prone soluble DPP4 might not be primarily responsible for decreased plasma concentrations, due to the circumstance that declines are not associated with the extent of resected liver volume.

We next evaluated for a possible association between bile acid, IL-6 or circulating DPP4 levels and the plasma GLP-1 and GLP-2 concentrations in the perioperative course. Preoperatively, a weak correlation of bile acids could be observed for both, GLP-1 and GLP-2. Postoperatively IL-6 levels showed a more pronounced association with GLP-1 and GLP-2 concentrations. GLP-2 levels on the other hand seemed to correlate at any timepoint with IL-6, even more distinct postoperatively. DPP4 showed no association with either GLP-1 or GLP-2 at any timepoint. (Table 2, Supplementary Materials & Methods Fig. S4).

PHLF and the GLP-1/GLP-2 ratio are associated with distinct postoperative plasma lipid dynamics

Given previous evidence, that lipid metabolism is altered during liver regeneration we aimed to study perioperative changes of clinical routine lipid parameters in liver regeneration.

Comparing baseline lipid metabolism parameters preoperatively between patients with and without PHLF, we found decreased levels of HDLc and ApoA1 in PHLF. Triglycerides, cholesterol, LDLc and ApoB level were seen unaltered. In both groups all lipid parameters declined postoperatively (Fig. 4a + d + g + j + m + p, Supplementary Materials & Methods Fig. S5).

Perioperative plasma lipid parameter dynamics in PHLF. Graphs and inserted columns with error-bars depict perioperative plasma concentration dynamics and AUC comparison for triglycerides (a), total cholesterol (d), LDLc (g), ApoB (j), HDLc (m) and ApoA1 (p). Violine blots represent fold-changes of plasma lipid parameter after 5 postoperative days, (b, e, h, k, n, q); Delineated are associations of postoperative changes of plasma lipid parameters with the postoperative alterations of GLP1/GLP-2 ratios in representative groups of most extreme postoperative change of the GLP-1/GLP-2 ratio (GLP-1/GLP-twofold-change low/high, n = 10, respectively) (c, f, i, l, o, r); horizontal and dotted lines for means and quartiles; Mann–Whitney-U test; LDLc low-density lipoprotein cholesterol, ApoB apolipoprotein B, HDLc high- density lipoprotein cholesterol, ApoA1 apolipoprotein A1; *p < 0.05; **p < 0.005; ***p < 0.0005.

When examining changes in plasma lipid levels until POD5, expressed as ratios to illustrate alterations relative to the baseline levels, only triglycerides and total cholesterol appeared to significantly decrease in PHLF. This decrease is presumably due to either increased consumption or exhausted mobilization/production in PHLF (Fig. 4b + e + h + k + n + q). Examining the time course, delineated as AUC, we did not observe any differences in triglyceride levels within the first 5 days postoperatively (Fig. 4a, Supplementary Materials & Methods Table S2).

Is has been reported previously, that GLP-1 and GLP-2 exhibit inverse effects on plasma lipid levels. To evaluate the balance of GLP-1 and GLP-2 on plasma lipid parameters, we integrated patients, who displayed the most extreme postoperative changes in the GLP-1/GLP-2 ratio, into a GLP-1/GLP-2 “high” and “low” group, and compared postoperative changes of lipid parameters, in order to delineate the combination of GLP effects on plasma lipid parameters postoperatively. We saw, that within patients who had a more pronounced postoperative decrease in the GLP-1/GLP-2 ratio, a significant association with a stronger decrease of lipid parameter (except triglycerides) could be observed (Fig. 4c + f + i + l + o + r). Given the fact that GLP-1 has a lipid lowering and GLP-2 a lipid increasing effect, these observations suggest a postoperative decline of GLP-1 and/or increase of GLP-2 levels may reflect a compensatory mechanism to maintain lipid homeostasis during liver regeneration. In patients with PHLF, postoperative GLP-2 AUCs trended to correlate with triglyceride AUCs (R = 0.750) and inversely with total cholesterol AUCs (R = − 0.821) (Supplementary Materials & Methods Table S5), suggesting that up to a certain point in PHLF, plasma lipid concentrations are linked to GLP-2 plasma levels.