Intratumoral PKM2 expression predicts benefit from adjuvant FOLFOX in stage II–III colorectal cancer

Introduction

Colorectal cancer is among the most commonly diagnosed malignancies worldwide and remains a leading cause of cancer-related mortality (1). Most cancer deaths are attributable to the progressive growth of metastatic lesions in secondary organs. Even among patients with non-metastatic colorectal cancer who undergo curative-intent resection, a substantial proportion (14–34%) later develop distant metastases during follow-up (2). Accordingly, current treatment paradigms recommend adjuvant chemotherapy for patients with high-risk, non-metastatic disease to eradicate hidden micrometastasis and improve survival.

In routine practice, the decision to administer adjuvant cytotoxic chemotherapy is largely based on estimated prognosis. Patients with adverse clinicopathological features—such as advanced tumor-node-metastasis (TNM) stage (e.g., T4 and/or lymph node positivity), poor differentiation, lymphovascular invasion, tumor budding, bowel obstruction, or perforation—are commonly considered appropriate candidates because their higher baseline risk can justify the potential harms associated with treatment toxicity (3). However, these factors are primarily prognostic; they do not necessarily identify a subgroup in which chemotherapy is intrinsically more effective. Under a proportional risk reduction framework, a similar relative treatment effect can translate into a larger absolute benefit in higher-risk patients, which supports risk-benefit-based treatment selection but does not establish a predictive biomarker. Therefore, identifying biomarkers that can predict differential benefit from specific adjuvant regimens could meaningfully refine patient selection and potentially reduce exposure to unnecessary toxicity.

Pyruvate kinase is a key enzyme in glycolysis, and the M2 isoform [pyruvate kinase M2 (PKM2)] is frequently expressed in malignant tissues and has been implicated in aerobic glycolysis and tumor proliferation (4,5). Because cytotoxic chemotherapy preferentially targets rapidly proliferating cells, intratumoral PKM2 expression has been proposed as a candidate biomarker for chemotherapy responsiveness (6). Prior experimental and clinical studies have suggested associations between PKM2 and chemosensitivity or chemoresistance across multiple tumor types (7-9), including colorectal cancer (10,11), and regimen-specific effects have been reported—such as links between lower PKM2 expression and oxaliplatin resistance in some settings (10), contrasted with findings implicating PKM2 upregulation in reduced sensitivity to fluoropyrimidines in others (11). These seemingly discordant observations support the hypothesis that PKM2 may function as a regimen-dependent marker rather than a universal indicator of chemotherapy benefit.

Importantly, establishing PKM2 as a predictive biomarker requires demonstrating that treatment effects differ by PKM2 status, ideally through formal treatment-by-biomarker interaction testing in the presence of an appropriate comparator group (12). Many previous studies have been limited by the absence of an untreated reference group and/or by not formally evaluating interaction, making it difficult to distinguish predictive effects from purely prognostic associations (10,11).

In this study, we aimed to evaluate whether intratumoral PKM2 expression predicts the effectiveness of adjuvant chemotherapy regimens in stage II and III colorectal cancer, with particular focus on oxaliplatin-based combination chemotherapy (FOLFOX). We compared outcomes across patients receiving no adjuvant chemotherapy (NoAC), fluoropyrimidine monotherapy (FP), and FOLFOX, and assessed predictive value by testing treatment-by-marker interactions. To facilitate unbiased comparisons between non-randomized treatment groups, we applied energy-based covariate balancing weights to balance measured baseline covariates and estimate average treatment effects (13). We present this article in accordance with the STROBE reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0024/rc).

Methods

Patients

This is a single-center, retrospective cohort study. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of GangNeung Asan Hospital (GNAH; No. 2024-05-015) and individual consent for this retrospective analysis was waived. Consecutive patients with resectable stage II and III colorectal cancer who had undergone upfront colorectal surgery between January 2008 and December 2012 were included. Patient accrual began in 2008 because FOLFOX was incorporated into routine clinical practice as an adjuvant regimen at our institution from this time onward. Because this was an exploratory, hypothesis-generating study, no formal a priori sample size calculation was performed; instead, we predefined a 5-year accrual window and included all consecutive eligible patients during this period. Patients were eligible regardless of whether they received adjuvant chemotherapy. We excluded patients aged >80 years; those with an overt synchronous second primary malignancy at the time of colorectal cancer diagnosis; and those with unassessable PKM2 immunohistochemical staining due to cautery artifact, tissue fragmentation, or incorrect tissue orientation.

Stage II disease was classified as low-risk only if all of the following criteria were met: non-T4 tumor, ≥12 examined lymph nodes, no bowel obstruction, no perforation, well or moderately differentiated histology, no lymphovascular invasion, no perineural invasion, and a negative resection margin; patients not meeting all of these criteria were conservatively classified as high-risk (14).

Clinicopathological data and outcome

Demographic and clinicopathological data were extracted from medical records by the authors, who were blinded to outcomes at the time of data abstraction. The following variables were collected: age at surgery, sex, bowel obstruction, perforation, pathological TNM stage (including T and N categories), lymphovascular invasion, perineural invasion, histologic grade, resection margin status, microsatellite instability or mismatch repair (MMR) status, diabetes mellitus (DM), hypertension, receipt of adjuvant chemotherapy, and (if applicable) chemotherapy regimen.

Bowel obstruction was defined as cancer-related luminal obstruction requiring emergency colectomy, diverting ostomy, or stenting. Perforation was assessed clinically rather than based solely on surgical pathology findings. Histologic grade was dichotomized based on gland formation, with tumors showing <50% gland formation classified as poorly differentiated. Chemotherapy regimens were categorized by agent(s) as FP or FOLFOX. The FP regimens included 5-fluorouracil (5-FU) plus leucovorin, capecitabine, doxifluridine, and tegafur/uracil. For comparative analyses, patients were grouped into NoAC, FP, and FOLFOX.

The primary endpoint was overall survival (OS), and the secondary endpoint was recurrence-free survival (RFS) (15). RFS was defined as the interval from the date of surgery to the date of first radiologic or histologic evidence of recurrence or death, whichever occurred first. OS was defined as the interval from the date of surgery to death from any cause. Patients without an event were censored at the date of last known follow-up, thereby addressing loss to follow-up through right-censoring. To minimize the influence of non-cancer-related deaths, follow-up was administratively censored at 10 years.

Assessment of PKM2 expression

Colorectal adenocarcinoma tissue specimens were formalin-fixed and paraffin-embedded. Tissue microarrays were constructed using a tissue-arraying instrument (Quick-Ray; Unitma Co., Ltd., Seoul, Republic of Korea). Representative tumor areas were identified and marked on hematoxylin and eosin-stained slides, and the corresponding paraffin blocks were sampled. From each donor block, a 2-mm-diameter core was punched and transferred into a recipient block. Cores were arrayed in duplicate to minimize tissue loss. Sections (4 µm thick) were cut from the tissue microarray blocks for immunohistochemical staining.

Immunohistochemistry for PKM2 (ab137791; Abcam, Cambridge, UK; 1:500) was performed on tissue microarray sections. Staining was performed using either a Bond-Max automated immunostainer (Leica Biosystems, Newcastle, UK) or a Ventana Benchmark automated staining system (Ventana Medical Systems, Tucson, AZ, USA), according to the manufacturers’ instructions. Testicular tissue served as a positive control, and negative controls were prepared by omitting the primary antibody.

PKM2 staining was semi-quantitatively assessed based on cytoplasmic staining intensity and the percentage of positive tumor cells. Staining intensity was graded as 0 (negative or equivocally weak), 1 (unequivocally weak/mild), 2 (moderate), or 3 (strong), using normal mucosa as the reference for strong staining (Figure 1). The proportion of positive cells was graded as 0 (<5%), 1 (5–25%), 2 (26–50%), or 3 (>51%). The final immunohistochemical score was calculated by multiplying the intensity and proportion scores. PKM2 positivity was defined as a final score ≥4, as reported previously (16).

Figure 1 Immunohistochemical staining showing PKM2 expression at ×100 magnification. (A) No staining. (B) 1+ indicates unequivocally weak-to-mild cytoplasmic staining. (C) 2+ indicates moderate cytoplasmic staining. (D) 3+ indicates strong cytoplasmic staining. PKM2, pyruvate kinase M2.

Statistical analysis

Energy balancing weights were used to balance baseline covariate distributions across treatment groups and to estimate average treatment effects in this observational cohort (13). This approach directly targets covariate balance and reduces reliance on correct specification of a treatment assignment model (17).

Weights were computed to balance the following covariates across treatment groups: age at diagnosis, sex, bowel obstruction, perforation, pathological T and N categories, lymphovascular invasion, tumor grade, DM, and hypertension. DM and hypertension were included as proxies for non-cancer comorbidity that could influence survival. Because there were only two cases with positive resection margins (n=2), margin status was not included in the weighting model. Covariate balance was assessed using absolute standardized mean differences, with values >0.1 considered indicative of meaningful imbalance.

Weighted Kaplan-Meier curves were generated for each treatment group. Weighted Cox proportional hazards models were used to estimate hazard ratios (HRs). Treatment-by-marker interaction terms were included in weighted Cox models to evaluate whether treatment effectiveness differed by PKM2 status (12).

Prespecified subgroup analyses were performed in a clinically relevant population after excluding low-risk stage II disease and stage II tumors with deficient MMR (dMMR), because adjuvant chemotherapy is generally not recommended for these subgroups in contemporary practice guidelines (14).

Missing values for DM and hypertension were imputed using multiple imputation by chained equations (18). Two-sided P values <0.05 were considered statistically significant. All analyses were conducted using R (version 4.5.1) (19).

Results

Patient characteristics

A total of 234 patients were screened for eligibility during the 2008–2012 study period. Of these, 193 patients were included in the final analysis after excluding 41 patients (Figure 2). Ninety-eight patients (51%) were PKM2-positive and 95 were PKM2-negative. Among PKM2-positive patients, 26 (27%), 49 (50%), and 23 (23%) received NoAC, FP, and FOLFOX, respectively. Among PKM2-negative patients, 22 (23%), 44 (46%), and 29 (31%) received NoAC, FP, and FOLFOX, respectively.

Figure 2 Study population. FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; NoAC, no adjuvant chemotherapy; PKM2, pyruvate kinase M2.

Baseline demographic and clinicopathological characteristics according to PKM2 status are summarized in Table 1. The prevalence of dMMR tended to be higher in PKM2-positive patients than in PKM2-negative patients (15.3% vs. 6.3%), although the difference did not reach statistical significance (P=0.08). PKM2 expression was not significantly associated with any other baseline characteristic.

The median follow-up duration was 11.8 years in the PKM2-positive group and 11.3 years in the PKM2-negative group.

Association between the PKM2 status and the effectiveness of different chemotherapy regimens

All covariates achieved absolute standardized mean differences ≤0.1 after weighting, suggesting no meaningful residual imbalance (Figure 3).

Figure 3 Love plot of covariate balance before and after weighting. Absolute standardized mean differences are shown for baseline covariates before (unadjusted) and after adjustment using energy balancing weights for each pairwise treatment comparison (FP vs. NoAC, FOLFOX vs. NoAC, and FOLFOX vs. FP). After weighting, no covariate exceeded an absolute standardized mean difference of 0.1. DM, diabetes mellitus; FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; HTN, hypertension; LVI, lymphovascular invasion; M, male; N, node; NoAC, no adjuvant chemotherapy; T, tumor; WMD, well or moderately differentiated.

Covariate-adjusted Kaplan-Meier curves were compared among NoAC, FP, and FOLFOX after weighting, stratified by PKM2 status (Figure 4). In the overall weighted cohort, FOLFOX showed no clear visual advantage over FP for either OS or RFS. In the PKM2-positive group, however, the FOLFOX curve consistently showed the most favorable separation for both OS and RFS, whereas outcomes with FP were intermediate and outcomes with NoAC were least favorable. In contrast, in the PKM2-negative group, the FOLFOX curves showed little separation from NoAC and appeared inferior to FP, suggesting qualitative effect modification of oxaliplatin-based therapy by PKM2 status. These patterns motivated subsequent formal estimation of regimen-specific HRs and treatment-by-marker interaction effects using weighted Cox proportional hazards models.

Figure 4 Weighted Kaplan-Meier curves by PKM2 status and adjuvant chemotherapy regimen. Weighted Kaplan-Meier curves for overall survival (A) and recurrence-free survival (B) are shown for all patients (left), patients with PKM2-positive tumors (middle), and patients with PKM2-negative tumors (right). Curves are weighted using energy balancing weights to improve comparability across treatment groups. Numbers at risk (raw counts) are presented below each panel. FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; NoAC, no adjuvant chemotherapy; PKM2, pyruvate kinase M2.

Accordingly, weighted Cox proportional hazards models incorporating treatment-by-marker interaction terms were fitted to formally quantify regimen-specific HRs by PKM2 status (Figure 5). For OS, FOLFOX was associated with substantially improved survival compared with NoAC in the PKM2-positive group [HR =0.10; 95% confidence interval (CI): 0.02–0.38; P=0.001], whereas no significant benefit was observed in the PKM2-negative group (HR =0.77; 95% CI: 0.33–1.79; P=0.55), with a significant treatment-by-marker interaction (P=0.01). When FOLFOX was compared with FP, PKM2-positive tumors again showed a survival benefit with FOLFOX (HR =0.25; 95% CI: 0.07–0.96; P=0.044), while PKM2-negative tumors showed an opposite-direction trend favoring FP (HR =2.39; 95% CI: 0.80–7.18; P=0.12), and the interaction remained significant (P=0.01). For RFS, similar effect modification by PKM2 status was observed: FOLFOX was associated with improved outcomes vs. NoAC in the PKM2-positive group (HR =0.11; 95% CI: 0.04–0.35; P<0.001) but not in the PKM2-negative group (HR =0.64; 95% CI: 0.27–1.54; P=0.32) (interaction P=0.01), and FOLFOX was favored over FP only in the PKM2-positive group (HR =0.18; 95% CI: 0.05–0.64; P=0.008) with an opposite-direction trend in the PKM2-negative group (HR =1.76; 95% CI: 0.59–5.28; P=0.31) (interaction P=0.006).

Figure 5 PKM2 modifies the association between adjuvant chemotherapy regimen and outcomes in weighted Cox models. HRs and 95% CIs for overall survival and recurrence-free survival are shown for FP vs. NoAC, FOLFOX vs. NoAC, and FOLFOX vs. FP, stratified by intratumoral PKM2 status. Estimates were obtained from weighted Cox proportional hazards models using energy balancing weights, with treatment-by-marker interaction P values shown for each regimen comparison. Squares indicate point estimates and horizontal lines indicate 95% CIs; arrows denote CIs extending beyond the plotted range. Treatment and control columns indicate unweighted sample sizes. CI, confidence interval; FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; HR, hazard ratio; NoAC, no adjuvant chemotherapy; PKM2, pyruvate kinase M2.

Subgroup analysis

To evaluate the robustness and clinical applicability of the primary findings, prespecified subgroup analyses were performed in a clinically relevant population excluding low-risk stage II disease and stage II tumors with dMMR. After weighting, covariate balance remained acceptable across treatment groups in this restricted cohort (absolute standardized mean differences ≤0.1) (Figure 6).

Figure 6 Covariate balance after weighting in the prespecified subgroup analysis. Love plots show absolute standardized mean differences for baseline covariates before (unadjusted) and after adjustment using energy balancing weights for each pairwise treatment comparison (FP vs. NoAC, FOLFOX vs. NoAC, and FOLFOX vs. FP) in the prespecified subgroup excluding low-risk stage II disease and stage II tumors with deficient mismatch repair. The dashed vertical line indicates an absolute standardized mean difference of 0.1. DM, diabetes mellitus; FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; HTN, hypertension; LVI, lymphovascular invasion; M, male; N, node; NoAC, no adjuvant chemotherapy; T, tumor; WMD, well or moderately differentiated.

In this subgroup, PKM2 status continued to modify the association between oxaliplatin-based therapy and outcomes (Figure 7). For OS, FOLFOX was associated with improved survival compared with NoAC in PKM2-positive tumors (HR =0.09; 95% CI: 0.02–0.35; P<0.001), whereas no significant benefit was observed in PKM2-negative tumors (HR =0.75; 95% CI: 0.32–1.73; P=0.49), with a significant treatment-by-marker interaction (P=0.01). Similarly, when FOLFOX was compared with FP, outcomes favored FOLFOX in PKM2-positive tumors (HR =0.23; 95% CI: 0.06–0.87; P=0.03) but showed an opposite-direction trend in PKM2-negative tumors (HR =2.15; 95% CI: 0.73–6.31; P=0.16), with a significant interaction (P=0.01).

Figure 7 Subgroup analysis in clinically relevant stage II/III disease. PKM2 modifies the association between adjuvant regimen and outcomes in weighted Cox models. HRs and 95% CIs for overall survival and recurrence-free survival are shown for FP vs. NoAC, FOLFOX vs. NoAC, and FOLFOX vs. FP, stratified by intratumoral PKM2 status, in the prespecified subgroup excluding low-risk stage II disease and stage II tumors with deficient mismatch repair. Estimates were obtained from weighted Cox proportional hazards models using energy balancing weights, with treatment-by-marker interaction P values shown for each regimen comparison. Treatment and control columns indicate unweighted sample sizes. Squares indicate point estimates and horizontal lines indicate 95% CIs; arrows denote CIs extending beyond the plotted range. CI, confidence interval; FOLFOX, oxaliplatin-based combination chemotherapy; FP, fluoropyrimidine monotherapy; HR, hazard ratio; NoAC, no adjuvant chemotherapy; PKM2, pyruvate kinase M2.

For RFS, consistent effect modification was observed: FOLFOX was associated with improved RFS vs. NoAC in PKM2-positive tumors (HR =0.11; 95% CI: 0.03–0.34; P<0.001) but not in PKM2-negative tumors (HR =0.63; 95% CI: 0.26–1.51; P=0.29) (interaction P=0.01). When compared with FP, FOLFOX was favored only in PKM2-positive tumors (HR =0.16; 95% CI: 0.05–0.55; P=0.004), whereas PKM2-negative tumors again showed an opposite-direction trend (HR =1.60; 95% CI: 0.55–4.66; P=0.38) (interaction P=0.005). These subgroup results were concordant with the primary analysis, supporting intratumoral PKM2 expression as a regimen-specific predictive biomarker for benefit from adjuvant FOLFOX.

Discussion

A biomarker is considered predictive of the effectiveness of a specific treatment when the treatment effect differs according to biomarker status (12). Predictive biomarkers can inform clinical decision-making by identifying patients who are more likely to benefit from a particular regimen, thereby enabling more personalized treatment strategies. In this study, intratumoral PKM2 expression was differentially associated with the effectiveness of adjuvant chemotherapy regimens in patients with resectable stage II and III colorectal cancer. Specifically, the survival benefit of FOLFOX compared with FP or NoAC was confined to PKM2-positive tumors, whereas this differential effectiveness was not observed in PKM2-negative tumors. Notably, significant treatment-by-marker interactions were observed for comparisons involving FOLFOX, supporting PKM2 status as a predictive biomarker for the effectiveness of adjuvant FOLFOX.

A key observation in our study was regimen-dependent heterogeneity between FOLFOX and FP according to PKM2 status. Because FOLFOX contains a fluoropyrimidine backbone, this pattern may reflect differences in the incremental benefit of oxaliplatin beyond fluoropyrimidine therapy. Prior studies reported downregulation of PKM2 in oxaliplatin-resistant colorectal cancer cell lines and an association between low PKM2 expression and poorer response to oxaliplatin-based chemotherapy (10,20), which is consistent with our finding that PKM2 positivity predicted greater benefit from FOLFOX (Figures 5,7). In contrast, PKM2 upregulation has been reported in 5-FU-resistant models and in patients with poor response to 5-FU (11), aligning with our observation of less favorable outcomes with FP in PKM2-positive tumors (Figures 5,7). However, many prior studies lacked an appropriate comparator group, limiting formal evaluation of treatment-by-marker interactions. By including patients treated with FOLFOX, FP, and NoAC, our study enabled regimen comparisons within PKM2 strata and formal interaction testing, further supporting PKM2 as a predictive biomarker for benefit from adjuvant FOLFOX.

A potential biological explanation for the regimen-dependent predictive value of PKM2 involves cancer metabolic reprogramming. The Warburg effect, characterized by preferential reliance on aerobic glycolysis despite adequate oxygen, is closely linked to proliferative biology (21,22), and PKM2 is a key glycolytic regulator implicated in tumor growth and metabolic adaptation (6). In colorectal cancer, APC loss has been reported to promote a Warburg-like phenotype through increased PKM2 transcription (23), linking a canonical driver of colorectal tumorigenesis to PKM2-mediated metabolic reprogramming (4,24). Oxaliplatin exerts cytotoxicity mainly through formation of DNA adducts and crosslinks that disrupt DNA replication and trigger apoptosis, whereas PKM2 may influence cellular tolerance to DNA damage and oxidative stress through metabolic and redox regulation (20,25). Consistent with this mechanism, prior colorectal cancer studies have linked PKM2 expression or regulation to oxaliplatin sensitivity and resistance (10,26). More broadly, recent molecular prediction efforts support the feasibility of biomarker-based stratification of oxaliplatin benefit (27,28), and future studies may further refine this approach in a multi-omic framework that integrates PKM2 with other biomarkers, including ctDNA-based assessment of molecular residual disease (29).

The clinical implication of the present study is that PKM2 status may help guide individualized selection of adjuvant chemotherapy. Current recommendations for adjuvant therapy in resected stage II/III colorectal cancer are largely based on a proportional risk reduction framework (30), in which more intensive regimens such as FOLFOX are preferentially offered to patients with poorer expected prognosis because the larger baseline risk translates into greater absolute benefit that may outweigh the added toxicity of oxaliplatin in this population. However, our findings suggest that the benefit of FOLFOX may not be uniform across all patients deemed high-risk by clinicopathological features; rather, benefit may be largely confined to PKM2-positive tumors. Consequently, administering oxaliplatin-containing adjuvant therapy to patients with PKM2-negative tumors may expose them to additional toxicity without a clear survival benefit. In such cases, alternative strategies—including omission of oxaliplatin with use of FP—may be more appropriate, although this hypothesis warrants prospective validation in future studies.

Because treatment was not randomized, residual confounding is a central concern when interpreting regimen-specific effects. To mitigate confounding by measured covariates, we applied energy balancing weights and confirmed good post-weighting balance across treatment groups. Although inverse probability of treatment weighting based on a propensity score model is commonly used for covariate adjustment, model-based weights can be sensitive to how the treatment assignment model is specified and may require iterative refinement to achieve adequate balance (17). In contrast, energy balancing weights are explicitly constructed to directly target distributional balance—using a distributional distance measure—thereby reducing reliance on correct specification of a parametric treatment assignment model (13). In this context, energy balancing weights enabled estimation of population-average (marginal) treatment effects that are conceptually aligned with the types of comparisons typically targeted in randomized trials. Moreover, this framework provided a coherent basis for evaluating effect modification by intratumoral PKM2 expression: regimen-specific HRs could be estimated within PKM2 strata, and treatment-by-marker interactions could be formally tested using weighted Cox proportional hazards models. In our study, covariate balance diagnostics demonstrated consistently low post-weighting standardized mean differences, and the resulting weighted analyses yielded a clear, internally consistent pattern of regimen-dependent benefit, in which the effectiveness of oxaliplatin-based therapy differed substantially according to PKM2 status.

This study has several limitations. First, because of the retrospective design, detailed measures of performance status and comorbidity were not consistently available and could not be fully incorporated into the analyses; diabetes and hypertension were therefore used as proxy variables, and patients with overt synchronous double primary malignancies were excluded. Second, although energy balancing weights achieved good balance for measured covariates, weighting approaches cannot account for unmeasured confounding and may be sensitive to limited overlap and weight variability. In particular, the overall weighted Kaplan-Meier curves showed no clear advantage of FOLFOX over FP and, in some panels, FOLFOX appeared less favorable. Because oxaliplatin-containing adjuvant therapy is more likely to be selected for patients with poor baseline prognosis, this pattern may reflect residual confounding by indication despite weighting. Third, some estimates—particularly in PKM2-negative tumors—were imprecise, with wide CIs, and require confirmation in larger, prospective cohorts. Fourth, because this was a single-center study, external validation in independent, multi-center cohorts is warranted, consistent with guidance to discuss generalizability (external validity) in observational research reports. Finally, this study did not investigate mechanistic pathways underlying the regimen-dependent predictive value of PKM2, and further work is needed to clarify the biological basis of these findings.

Conclusions

Intratumoral PKM2 expression may predict benefit from adjuvant FOLFOX in patients with stage II/III colorectal cancer. In our cohort, the survival benefit of adjuvant FOLFOX was observed in the PKM2-positive group but not in the PKM2-negative group. Thus, PKM2 status may help guide selection of an optimal adjuvant chemotherapy regimen and warrants prospective validation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0024/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0024/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0024/prf

Funding: This work was supported by the Medical Research Promotion Program, through the GangNeung Asan Hospital, funded by the Asan Foundation (Nos. 2023II0007 and 2025II0007).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0024/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of GangNeung Asan Hospital (GNAH; No. 2024-05-015) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Morgan E, Arnold M, Gini A, et al. Global burden of colorectal cancer in 2020 and 2040: incidence and mortality estimates from GLOBOCAN. Gut 2023;72:338-44. [Crossref] [PubMed]
2. van Gestel YR, de Hingh IH, van Herk-Sukel MP, et al. Patterns of metachronous metastases after curative treatment of colorectal cancer. Cancer Epidemiol 2014;38:448-54. [Crossref] [PubMed]
3. Benson AB, Venook AP, Al-Hawary MM, et al. Colon Cancer, Version 2.2021, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2021;19:329-59. [Crossref] [PubMed]
4. Dayton TL, Jacks T, Vander Heiden MG. PKM2, cancer metabolism, and the road ahead. EMBO Rep 2016;17:1721-30. [Crossref] [PubMed]
5. Puckett DL, Alquraishi M, Chowanadisai W, et al. The Role of PKM2 in Metabolic Reprogramming: Insights into the Regulatory Roles of Non-Coding RNAs. Int J Mol Sci 2021;22:1171. [Crossref] [PubMed]
6. Zhao Y, Butler EB, Tan M. Targeting cellular metabolism to improve cancer therapeutics. Cell Death Dis 2013;4:e532. [Crossref] [PubMed]
7. Zhu H, Wu J, Zhang W, et al. PKM2 enhances chemosensitivity to cisplatin through interaction with the mTOR pathway in cervical cancer. Sci Rep 2016;6:30788. [Crossref] [PubMed]
8. Wang Y, Zhao H, Zhao P, et al. Targeting PKM2 promotes chemosensitivity of breast cancer cells in vitro and in vivo. Cancer Biomark 2021;32:221-30. [Crossref] [PubMed]
9. Papadaki C, Sfakianaki M, Lagoudaki E, et al. PKM2 as a biomarker for chemosensitivity to front-line platinum-based chemotherapy in patients with metastatic non-small-cell lung cancer. Br J Cancer 2014;111:1757-64. [Crossref] [PubMed]
10. Martinez-Balibrea E, Plasencia C, Ginés A, et al. A proteomic approach links decreased pyruvate kinase M2 expression to oxaliplatin resistance in patients with colorectal cancer and in human cell lines. Mol Cancer Ther 2009;8:771-8. [Crossref] [PubMed]
11. Shin YK, Yoo BC, Hong YS, et al. Upregulation of glycolytic enzymes in proteins secreted from human colon cancer cells with 5-fluorouracil resistance. Electrophoresis 2009;30:2182-92. [Crossref] [PubMed]
12. Ballman KV. Biomarker: Predictive or Prognostic? J Clin Oncol 2015;33:3968-71. [Crossref] [PubMed]
13. Huling JD, Mak S. Energy balancing of covariate distributions. Journal of Causal Inference 2024;12:20220029.
14. Baxter NN, Kennedy EB, Bergsland E, et al. Adjuvant Therapy for Stage II Colon Cancer: ASCO Guideline Update. J Clin Oncol 2022;40:892-910. [Crossref] [PubMed]
15. Birgisson H, Wallin U, Holmberg L, et al. Survival endpoints in colorectal cancer and the effect of second primary other cancer on disease free survival. BMC Cancer 2011;11:438. [Crossref] [PubMed]
16. Cui R, Shi XY. Expression of pyruvate kinase M2 in human colorectal cancer and its prognostic value. Int J Clin Exp Pathol 2015;8:11393-9.
17. Kang JDY, Schafer JL. Demystifying Double Robustness: A Comparison of Alternative Strategies for Estimating a Population Mean from Incomplete Data. Statist Sci 2007;22:523-39.
18. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. Journal of Statistical Software 2011;45:1-67.
19. R Core Team. R: A language and environment for statistical computing. R v4.3.2. Vienna: R Foundation for Statistical Computing; 2023. Available online: https://www.R-project.org/
20. Ginés A, Bystrup S, Ruiz de Porras V, et al. PKM2 Subcellular Localization Is Involved in Oxaliplatin Resistance Acquisition in HT29 Human Colorectal Cancer Cell Lines. PLoS One 2015;10:e0123830. [Crossref] [PubMed]
21. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646-74. [Crossref] [PubMed]
22. Liberti MV, Locasale JW. The Warburg Effect: How Does it Benefit Cancer Cells? Trends Biochem Sci 2016;41:211-8. [Crossref] [PubMed]
23. Cha PH, Hwang JH, Kwak DK, et al. APC loss induces Warburg effect via increased PKM2 transcription in colorectal cancer. Br J Cancer 2021;124:634-44. [Crossref] [PubMed]
24. Mazurek S. Pyruvate kinase type M2: a key regulator of the metabolic budget system in tumor cells. Int J Biochem Cell Biol 2011;43:969-80. [Crossref] [PubMed]
25. Seetharam R, Sood A, Goel S. Oxaliplatin: pre-clinical perspectives on the mechanisms of action, response and resistance. Ecancermedicalscience 2009;3:153. [Crossref] [PubMed]
26. Sfakianaki M, Papadaki C, Tzardi M, et al. PKM2 Expression as Biomarker for Resistance to Oxaliplatin-Based Chemotherapy in Colorectal Cancer. Cancers (Basel) 2020;12:2058. [Crossref] [PubMed]
27. Abraham JP, Magee D, Cremolini C, et al. Clinical Validation of a Machine-learning-derived Signature Predictive of Outcomes from First-line Oxaliplatin-based Chemotherapy in Advanced Colorectal Cancer. Clin Cancer Res 2021;27:1174-83. [Crossref] [PubMed]
28. Chen L, Wang Y, Cai C, et al. Machine Learning Predicts Oxaliplatin Benefit in Early Colon Cancer. J Clin Oncol 2024;42:1520-30. [Crossref] [PubMed]
29. Tie J, Cohen JD, Lahouel K, et al. Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon Cancer. N Engl J Med 2022;386:2261-72. [Crossref] [PubMed]
30. Wang H, Boissel JP, Nony P. Revisiting the relationship between baseline risk and risk under treatment. Emerg Themes Epidemiol 2009;6:1. [Crossref] [PubMed]