This multicentric, retrospective study analyzed 240 patients with mRCC treated with CPI-based 1L therapy to evaluate the prognostic value of serum liver enzymes and LDH. High baseline levels of LDH and GGT were independently associated with shorter PFS and/or OS. Stratification by combined GGT/LDH risk groups further enhanced prognostic discrimination: patients with both markers elevated had significantly worse survival outcomes.
Treatment and response patterns
Patients in the study were treated with a range of CPI-based combination therapies, which resulted in varied outcomes with 4% CR, 44% PR and 32% SD. Only 20% experienced PD reflecting the efficacy of CPI-based therapies in mRCC. However, the objective response rate (CR + PR: 48%) was slightly worse than in the setting of clinical trials14,15,16,17, which can be expected in a real-world setting.
Baseline values of ALAT, ASAT, DRR, GGT and LDH were also examined in relation to treatment response. However, no significant differences in baseline levels of these markers were observed across the response groups. This is in contrast to the preceding study, where response to therapy (CR + PR) was more frequent in patients with low baseline levels of ALAT, ASAT, GGT and LDH12. Overall, this suggests that while these biomarkers are often implicated in liver function and systemic inflammation, they may not serve as reliable standalone biomarkers for response in this cohort, underlining the complexity of immune-related response mechanisms18,19.
Clinical parameters as prognostic factors for PFS and OS
Among the clinical parameters, IMDC risk category, nephrectomy status, presence of liver metastasis and type of combination therapy emerged as prognostic factors for PFS and/or OS, which aligns with findings from other real-world cohorts20,21,22. Patients with past nephrectomy exhibited improved survival, while poor IMDC risk was linked to increased overall mortality. Liver metastasis was significantly associated with reduced survival, corroborating its role as a negative prognostic factor in mRCC20,23,24. The presence of liver involvement often correlates with a more aggressive disease course and poorer response to CPI-based therapy20,24. Furthermore, the results also indicated that CPI + TKI combinations provided superior PFS compared to dual CPI therapies. Advantageous outcomes for treatments with CPI + TKI have also been reported in other real-world cohorts of mRCC20,21,22.
Baseline LDH and GGT as prognostic factors for PFS and OS
In the preceding monocentric study, high baseline levels of ALAT, ASAT, GGT and LDH were independently associated with a higher progression rate and increased overall mortality in patients with mRCC treated with CPI-based 1L therapy12. Particularly for OS, GGT and LDH displayed a higher prognostic value than the investigated clinical prognostic factors. Here, we could corroborate GGT and LDH as independent prognosticators in a larger, multicentric cohort. High LDH levels at baseline were significantly associated with worse PFS and OS outcomes, suggesting that LDH might be a reliable marker of tumor burden and systemic inflammation in patients with mRCC treated with CPI-based 1L therapies. Similarly, elevated GGT levels correlated with poor OS, further emphasizing the association of liver function markers with oncological outcomes in patients with mRCC undergoing CPI-based 1L therapy. In contrast, the prognostic value of ALAT and ASAT could not be confirmed in the present study. In accordance with the preceding study12, the DRR as the ASAT/ALAT ratio was without prognostic significance.
In accordance with our findings, high levels of GGT and LDH were also associated with worse survival in patients with mRCC receiving ≥2L nivolumab13,25,26 as well as in CPI-treated melanoma, lung cancer and hepatocellular cancer27,28,29,30. GGT and LDH are involved in cancer metabolism and thus might reflect tumor burden and systemic inflammation6,7,8. Via conversion of pyruvate to lactate, LDH restores NAD+ for ongoing aerobic glycolysis (Warburg effect), which provides sufficient energy and key nutrients for cancer cell proliferation7. GGT is involved in the metabolism of the antioxidant glutathione and can promote carcinogenesis and tumor progression6. Both GGT and LDH have also been shown to exert immunosuppressive effects8,31 and thus could influence efficacy of immunotherapy.
Notably, the combination of GGT and LDH risk groups enhanced their prognostic power. The presence of both high GGT and LDH levels was significantly and independently associated with a higher progression rate and increased overall mortality. Particularly for OS, the combined GGT/LDH score outperformed the prognostic value of clinico-pathological parameters, such as IMDC risk, and added additional prognostic value in multivariate models as indicated by higher C-indices. In accordance, prognostic models including GGT and LDH better stratified patients with advanced pancreatic ductal adenocarcinoma treated with chemotherapy32 and urothelial carcinoma of the upper urinary tract after radical nephroureterectomy33 regarding their prognosis. The combined elevation of GGT and LDH might reflect increased oxidative stress and a changed energy metabolism due to their role in glutathione metabolism and glycolysis, respectively6,7,8.
In subgroup analysis regarding therapy regimen, a significant association of the combined GGT/LDH score with prognosis could only be observed in the CPI + TKI treatment group, but not in the CPI + CPI subgroup. These observations may be explained by a combination of biological and statistical factors: The subgroup of patients treated with CPI + CPI was smaller than the CPI + TKI subgroup (72 vs. 162). Thus, statistical testing might be underpowered. On the other hand, we detailed above that LDH and GGT reflect metabolic aggressiveness and oxidative stress, which are particularly relevant for the efficacy of TKI-based regimens that target hypoxia-driven angiogenesis and tumor metabolism. The prognostic impact of combined GGT/LDH therefore appears more pronounced in CPI + TKI-treated patients than in those receiving CPI + CPI.
Furthermore, the combined GGT/LDH score showed prognostic potential for PFS in the subgroup with poor IMDC risk, whereas a prognostic association with OS could be demonstrated regardless of IMDC risk, albeit reaching statistical significance only in the subgroups with good and poor IMDC risk. In the IMDC good- and poor-risk categories, the GGT/LDH score is likely capturing either unexpectedly aggressive biology (good risk) or amplifying already pronounced metabolic dysregulation (poor risk), resulting in larger effect sizes. In contrast, intermediate-risk patients represent a more heterogeneous biological group in which the discriminative power of GGT and LDH is reduced. Taken together, these biological considerations, combined with differences in sample size and effect magnitude across subgroups, likely explain the observed pattern.
From a clinical perspective, the combined GGT/LDH score may serve as a simple, inexpensive and routinely available biomarker-based tool to support risk stratification. Patients presenting with both elevated GGT and LDH at the beginning of therapy could be recognized as having a particularly unfavorable prognosis and may benefit from intensified monitoring such as early imaging re-evaluation. Thereby, non-responders undergoing progression could potentially be identified earlier, and thus, therapy adjustments could be initiated without delay. Conversely, patients with low baseline levels of both markers might represent a subgroup with more favorable outcomes and could potentially be managed with standard follow-up intervals. Therefore, the GGT/LDH score could help refine personalized treatment strategies and improve patient counseling in the real-world management of mRCC.
Limitations
Owing to its retrospective design, our study is not devoid of limitations. There could be a selection bias due to the exclusion of patients with missing or insufficient values for the assessed laboratory parameters. However, only 18 (7%) patients were excluded from further analysis from the originally 258 screened patients (Supplementary Fig. S1). Furthermore, biomarker distributions did not differ significantly among study centers.
Another potential limitation could be inter-laboratory variability in GGT and LDH assays across participating centers. However, all laboratories were accredited and followed standardized quality control procedures, and no significant differences in baseline median values could be observed between study centers. Prognostic groupings for each parameter (high vs. low) were defined based on the overall cohort median, which guarantees comparable group sizes for statistical analyses and minimizes bias from site-specific differences. However, this approach may not reflect optimal clinical thresholds, and consequently, the optimal cutoff values for GGT and LDH need to be prospectively determined. Nevertheless, the median values of GGT (34.2 U/l) and LDH (222 U/l) concurred with those in the preceding smaller monocentric study12 and were similar to the cutoff values used in other studies investigating the prognostic impact of both parameters in CPI-treated cancer patients (GGT: 42–71 U/l, LDH: 196–264 U/l) 13,25–30.
Baseline values were sampled within the time frame of − 41 to + 7 days relative to treatment initiation, which could potentially introduce minor variability into the analyses as laboratory values may have changed during this interval. However, most values (n = 208, 86.7%) were obtained within the interval of -7 to + 7 days around treatment start, and thus, the influence of timing heterogeneity on the results is expected to be limited.
Furthermore, the absence of a centralized review of radiological imaging could introduce some variability in the determination of progression and response. Potential confounding factors (e.g., liver disease, alcohol use, concurrent medications), which could influence GGT and LDH levels, were not comprehensively recorded as well. Therefore, the findings should be interpreted with caution in the context of a retrospective multi-institutional study. Lastly, although the number of events per degree of freedom met commonly recommended thresholds for multivariable Cox regression, the OS model operated at the lower end of this range. Therefore, a minor risk of overfitting cannot be fully excluded.
Nevertheless, the present cohort included all available CPI-based 1L therapy options for mRCC and thus reflects the current therapeutic landscape for mRCC. Perspectively, baseline GGT and LDH levels could aid clinicians in risk stratification and therapeutic decision-making. A prospective validation in larger, independent cohorts is warranted to optimize the time frame for laboratory sampling, establish cutoff values and assess potential confounders, thereby confirming the utility of GGT and LDH as monitoring biomarkers in clinical practice.
