Study of phosphorylated ribosomal protein S6 (pS6) in the clinical outcomes of patients undergoing hepatectomy for metastatic colorectal cancer

Abstract

Introduction

Colorectal cancer (CRC) is one of the most prevalent malignancies worldwide, with up to 50% of patients developing metastatic disease, predominantly involving the liver. Hepatectomy remains the gold standard for resectable hepatic metastases; however, recurrence rates remain high and reliable prognostic biomarkers are lacking. Phosphorylated ribosomal protein S6 (pS6), a surrogate marker of PI3K/AKT/mTORC1 pathway activation, has been implicated in tumor aggressiveness across multiple cancer types, but its prognostic role after hepatectomy for metastatic colorectal cancer (mCRC) has not been established.

Objective

To evaluate the prognostic impact of pS6 expression and other molecular markers (BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS, and HSPA9) assessed by immunohistochemistry on tissue microarray (TMA) on the clinical outcomes of patients undergoing hepatectomy for metastatic colorectal cancer.

Materials and methods

A retrospective cohort study was conducted including 64 patients who underwent hepatectomy for mCRC between 2012 and 2022 at a high-complexity oncology center in Northeast Brazil. Molecular marker expression was evaluated by immunohistochemistry on TMA constructed from paraffin-embedded hepatic metastasis specimens. Disease-free survival (DFS) at 1, 3, and 5 years and overall survival (OS) were analyzed using Fisher’s exact test, Kaplan–Meier curves, and Cox regression multivariate analysis.

Results

Median OS was 23 months (95% CI = 18.33–27.66). Among all molecular markers assessed, only pS6 positivity (20.7% of cases) showed a statistically significant association with worse DFS at 1 year (p = 0.032), 3 years (p = 0.038), and 5 years (p = 0.018). A significant association was also identified between pS6 positivity and FUS positivity (p = 0.014). In multivariate analysis for DFS, pS6 positivity was an independent predictor of recurrence at 3 years (HRa = 9.710; p = 0.028) and 5 years (HRa = 11.857; p = 0.018), as were postoperative complications at 3 years (HRa = 8.671; p = 0.014) and 5 years (HRa = 6.912; p = 0.029). For overall survival, three independent predictors of mortality were identified: adjuvant chemotherapy (HRa = 0.052; p < 0.001), absence of 1-year DFS (HRa = 3.641; p = 0.001), and pS6 positivity (HRa = 4.547; p = 0.001).

Conclusion

Median overall survival was 23 months, reflecting the high mortality burden of mCRC after hepatectomy. Among all molecular markers evaluated, pS6 was the only one independently associated with worse disease-free survival and higher mortality risk, consistent with its role as a surrogate marker of PI3K/AKT/mTORC1 pathway activation, though a direct causal relationship cannot be established from the present data. A significant association between pS6 and FUS positivity suggests a possible convergence of oncogenic pathways warranting further investigation. Postoperative complications were independent predictors of recurrence, and adjuvant chemotherapy was the strongest independent predictor of improved overall survival. Given the retrospective single-center design and limited sample size, these findings are hypothesis-generating and require prospective multicenter validation before clinical application.

Introduction

Colorectal cancer (CRC) affects men and women proportionally, with 90% of cases occurring in individuals over 50 years of age [1, 2]. It is associated with aging, inadequate diet, increased risk factors (smoking and sedentary lifestyle) and maintains a high mortality rate [3]. A worrying 35% of patients present with metastasis at the time of diagnosis, and up to 50% of CRCs may progress to metastatic colorectal cancer (mCRC) [4]. These statistics are alarming because CRC progresses slowly and pre-neoplastic lesions are frequently detected and treated during colonoscopy [5].

Research in molecular oncology of CRC, including metastatic CRC, is advancing rapidly, leading to a better understanding of this complex disease. Vogelstein’s pioneering work in 1990 on adenoma-carcinoma sequencing provided initial insights into CRC carcinogenesis and mutations in the APC, KRAS, DCC, and P53 genes [6, 7]. Among the genes associated with CRC, the tumor suppressor gene P53 is frequently mutated [8]. Mutations in the BRAF gene, predominantly the V600E substitution, are also found in CRC and are associated with worse clinical outcomes [9]. Studies have demonstrated an interaction between the proto-oncogenes β- catenin and FUS, showing a decrease in intestinal crypts with the FUS protein expressed in the nuclei of CRC cells [10]. Ribonucleoprotein 6 (S6), a component of the 40 S ribosomal subunit, is a phosphorylation target and can be used to assess the activity of protein kinases, such as those of the RAF and RAS pathways [11]. Phosphorylated ribosomal protein S6 (pS6) regulates mRNA translation and is a surrogate marker for activated PI3K/AKT/mTORC1 signaling, a pathway implicated in many types of cancer (Fig. 1) [12].

Fig. 1.

Schematic diagram of the Ribonucleoprotein 6 (S6) regulatory pathway. Arrows indicate unidirectional regulation. Source: Yi YW et al. Int J Mol Sci. 2021;23(1):48 [12]. Adapted under CC BY 4.0 license. Blue arrow: activation; red arrow: inactivation. In blue: induction by p-RPS6 or reduction by RPS6-KD. In red: reduction by p-RPS6 or induction by RPS6-KD. Red square: target with inhibitors

HSPA9A (mortalin), a heat shock protein (HSP), is overexpressed in some types of cancer, including CRC, and is correlated with reduced survival [13]. Activation of the HER2 pathway is a significant mechanism of resistance to epidermal growth factor receptor (EGFR) therapy, a crucial treatment for several malignancies. Studies have demonstrated positive HER2 expression in CRC [14]. EGFR overexpression is more common in metastatic CRC, making EGFR a crucial therapeutic target [15].

The liver is the main site of metastasis for mCRC. Hepatectomy is the gold standard for resectable lesions. It has been increasingly preferred due to multimodal treatment approaches, parenchyma-preserving surgical techniques, and minimally invasive procedures, enabling curative treatment for a more significant number of patients [16, 17]. Despite advances in mCRC treatment, cure rates remain low. A better understanding of the pathways involved in cancer cells has facilitated the development of more targeted therapies [18].

The molecular markers evaluated in this study were selected a priori based on three complementary criteria: their established or emerging prognostic relevance in colorectal cancer reported in the literature, their biological representativeness of key oncogenic pathways implicated in tumor aggressiveness and treatment resistance, and the availability of validated antibodies in the institutional laboratory. Specifically, pS6 was selected as a surrogate marker of PI3K/AKT/mTORC1 pathway activation, a central mechanism of proliferation, apoptosis resistance, and chemotherapy resistance in mCRC [11, 12]. BRAF V600E was included given its well-documented association with poor prognosis in CRC and its interaction with downstream S6 phosphorylation [9, 19]. p53 was selected due to its frequent mutation in CRC and its pivotal role in the adenoma-carcinoma sequence [6, 8]. CERBB2 (HER2) was included in light of its emerging role as a therapeutic target and mechanism of EGFR resistance in mCRC [14, 20]. MSH6 and PMS2 were selected to characterize mismatch repair (MMR) status and microsatellite instability, which carry prognostic and predictive implications in CRC [15]. FUS was included based on evidence of its proto-oncogenic interaction with β-catenin in colorectal adenocarcinoma [10]. Finally, HSPA9 (mortalin) was selected due to its overexpression in CRC and its association with reduced survival through inhibition of p53 tumor suppressor activity [13]. Together, this panel was designed to capture complementary molecular dimensions of tumor biology relevant to the clinical outcomes of patients undergoing hepatectomy for mCRC.

This study aims to evaluate the influence of prognostic factors and molecular biology, analyzing BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS, HSPA9, and pS6 on the clinical outcomes of patients undergoing hepatectomy.

Materials and methods

Study design, sample selection, and ethical considerations

This retrospective, observational, and quantitative cohort study analyzed electronic medical records and paraffin-embedded liver tissue blocks from 64 patients who underwent hepatectomy for the treatment of metastatic colorectal cancer (mCRC) and met predefined inclusion and exclusion criteria. The study investigated the influence of prognostic factors and molecular markers (BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS, HSPA9, pS6) on the clinical outcomes of patients treated at a high-complexity oncology center (CACON) in Northeast Brazil, between January 2012 and December 2022.

The study followed the STROBE checklist guidelines [21] and was approved by the hospital’s Ethics Committee. All procedures adhered to Brazilian ethical guidelines for research involving human beings (Resolution CNS 466/12), maintaining patient confidentiality and following the principles of the Declaration of Helsinki and the Nuremberg Code. The principal investigator’s supervisor, the Ethics Committee, and the hospital administration approved the study protocol.

Inclusion and exclusion criteria

Patients who underwent hepatectomy for metastatic colorectal cancer between January 2012 and December 2022 and who had documented outpatient follow-up were included. Data were collected from electronic medical records and analysis of paraffin-embedded liver tissue blocks. Patients with incomplete follow-up data, those who underwent hepatectomy for metastases from other primary sites, those who underwent metastasectomy, those without archived tissue samples for slide review and tissue microarray (TMA) construction, those with insufficient tissue for immunohistochemistry, or those without residual tumor in the archived samples were excluded.

Collection and retrieval of data from paraffin blocks

Sociodemographic data included sex, age, primary tumor location (right colon, transverse colon, left colon, sigmoid colon, or rectum), presence of synchronous metastases, number of metastases, length of hospital stay, presence of postoperative complications (using the Clavien-Dindo classification), indication for neoadjuvant or adjuvant therapy (and type), priority of treatment approach (resection of the primary tumor followed by hepatectomy, initial hepatic approach, or simultaneous approach and the justification), disease-free survival and overall survival (from surgery to the last follow-up or death), outcome, recurrence (location, diagnosis, and treatment), and biomarker status (BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS, HSPA9, pS6).

TMA construction and immunohistochemical processing

Tumor areas were demarcated on histological slides using a Pilot^® pen. Two 1.0 mm tissue cylinders were obtained from each paraffin-embedded block, transferred to a 170-well TMA receptor block, and sectioned at 4 μm onto silanized slides. One slide was stained with hematoxylin and eosin to assess tumor representation. Antigen retrieval was performed using the Dako/Agilent PT Link system (pH 6.0, incubation for 60 min). Immunohistochemical staining was automated on the Autostainer Link 48 system (Dako/Agilent) using the following antibodies: BRAF V600E (ThermoFisher, clone RM8, 1:200 dilution), p53 (Dako/Agilent, clone DO-1, ready to use), CERBB2 (Dako/Agilent, clone 3B5, ready to use), MSH6 (Dako/Agilent, clone EP49, ready to use), PMS2 (Dako/Agilent, clone EP51, ready to use), FUS (Santa Cruz, clone 4H11, 1:100 dilution), HSPA9 (Santa Cruz, clone D-9, 1:100 dilution) and pS6 (Santa Cruz, clone 50ser235/236, 1:100 dilution). The incubation time for the primary antibody was 30 min, and the reaction was visualized with the diaminobenzidine (DAB) chromogen.

A pathologist, unaware of the clinical data, interpreted the results. The BRAF V600E mutation was classified as positive due to strong staining. The MSH6 and PMS2 markers were classified as positive or negative based on the presence or absence of expression. Positive staining was defined as the presence of staining in at least 1% of cells, regardless of intensity. Cases without staining of either the neoplasm or the adjacent stroma were considered inconclusive (absence of internal control). The HSPA9, FUS, pS6, and p53 markers were classified using the H-score. For p53, scores of 0 indicated null expression, 1 to 30 indicated weak/focal expression (non-mutated pattern), and > 30 indicated positive expression. The CERBB2 marker was classified as follows: 0 - absent; 1+ - weak in > 10% of cells; 2+ - moderate in > 10% of cells; 3+ - strong in > 10% of cells. All slides were evaluated in a single reading session by one experienced pathologist blinded to clinical outcomes. Internal controls were used to validate staining adequacy: for MSH6 and PMS2, cases in which neither neoplastic nor stromal cells showed positivity were classified as inconclusive and excluded; for the remaining markers, adjacent non-neoplastic tissue served as an internal reference for staining quality. Staining was performed on an automated platform (Autostainer Link 48, Dako/Agilent) under standardized conditions, which minimizes run-to-run variability. Regarding TMA sampling, two 1.0 mm cores per case were obtained, which may not fully capture intratumoral heterogeneity; this is an inherent limitation of the TMA approach and should be considered when interpreting marker expression results.

Disease-free survival was calculated as the time elapsed between diagnosis and the last follow-up visit or recurrence (location and therapeutic approach recorded). Overall survival was determined by reviewing medical records to identify postoperative complications and by actively monitoring patients to verify their outcomes. Death was confirmed by follow-up exceeding two years (patients who were debilitated at the last visit, lost to follow-up, under treatment, or with postoperative complications such as pulmonary metastases or stroke), death records, or contact with family members. Living patients were defined as those with recent follow-up (< 2 years), stable patients with follow-up > 2 years (confirmed by contact), and those who were discharged from the hospital.

Statistical analysis

Data are presented as absolute and percentage frequencies. Disease-free survival at 1, 3, and 5 years was analyzed using Fisher’s exact test or Pearson’s chi-square test. Kaplan-Meier curves for disease-free survival and overall survival were constructed, with time calculated as the difference between the date of surgery and the date of recurrence or the last follow-up. Analyses were performed using SPSS version 20.0 for Windows, with a 95% confidence interval and p < 0.05 considered significant (two-tailed). Receiver Operating Characteristic (ROC) curve analysis was used to determine cut-off points for continuous variables. Variables included in the multivariate Cox regression models were selected a priori based on established clinical relevance rather than univariate p-value thresholds, in order to avoid data-driven variable selection bias. The proportional hazards assumption was not formally tested given the small sample size; this represents an acknowledged limitation of the present analysis.

Results

Epidemiological data regarding sex, age, primary tumor, synchronous lesions, and number of liver lesions in patients with metastatic colorectal cancer after hepatectomy at 1, 3, and 5 years are described in Table 1.

Table 1.

Epidemiological data on sex, age, primary tumor, synchronous lesions, and number of liver lesions in patients with metastatic colorectal cancer after hepatectomy at 1, 3, and 5 years

	n (%)
Total	64 (100.0%)
Sex
Feminine	34 (53.1%)
Masculine	30 (46.9%)
Age
Under 40 years old	1 (1.6%)
40–60 years old	21 (32.8%)
> 60 years old	42 (65.6%)
Primary tumor
Right colon	9 (14.1%)
Transverse colon	1 (1.6%)
Left colon	8 (12.5%)
Sigmoid	18 (28.1%)
Rectum	28 (43.8%)
Synchronous lesions
No	27 (42.2%)
Yes	37 (57.8%)
Initial liver assessment	11 (17.2%)
Simultaneous approach	2 (3.1%)
Primary tumor followed by hepatectomy	27 (42.2%)
Number of liver lesions
Up to 3	51 (79.7%)
3–5	12 (18.8%)
> 5	1 (1.6%)

Source: Prepared by the author (2024) with epidemiological data expressed as absolute and percentage frequencies

In Table 1, we identified that the sample consisted of 64 patients with metastatic colorectal cancer who underwent hepatectomy. Of these patients, the majority were female (n = 34, 53.1%) and 30 were male (46.9%). Regarding age range, patients over 60 years of age predominated (n = 42, 65.6%). There were 21 patients (32.8%) between 40 and 60 years of age, and 1 patient under 40 years of age. The majority of patients had synchronous lesions, totaling 37 patients (57.8%), while 27 did not (42.2%). The most prevalent initial approach was resection of the primary tumor followed by hepatectomy, 27 (42.2%), followed by initial hepatic approach, with 11 patients (17.2%), and simultaneous approach in 2 patients (3.1%). Most patients had up to three lesions, 51 patients (79.7%), followed by 3 to 5 lesions, 12 patients (18.8%), and more than 5 lesions, 1 patient (1.6%).

Regarding length of hospital stay and postoperative complications, the data are described in Table 2.

Table 2.

Epidemiological data on length of hospital stay and postoperative complications after hepatectomy at 1, 3, and 5 years

	n (%)
Length of hospital stay
Up to 7 days	33 (91.7%)
7–15 days	1 (2.8%)
> 15 days	2 (5.6%)
Postoperative complications
No	21 (32.8%)
Yes	43 (67.2%)
I	9 (14.1%)
II	25 (39.1%)
III	2 (3.1%)
IV	2 (3.1%)
V	5 (7.8%)

Source: Prepared by the author (2024) with epidemiological data expressed as absolute and percentage frequencies

The predominant length of hospital stay was up to 7 days 33 (91.7%). Two-thirds of the sample presented postoperative complications, with 43 patients (67.2%): grade II was the most frequent complication (n = 25, 39.1%). The next most common was grade I (n = 9, 14.1%), grade III (n = 2, 3.1%), grade IV (n = 2, 3.1%) and grade V (n = 5, 7.8%) with 5 postoperative deaths (Table 2).

Neoadjuvant chemotherapy was used in 43 patients (67.2%), with the FLOX regimen being the most frequently described protocol in 32 patients (50.0%). Adjuvant chemotherapy was used in 57 patients (89.1%), with the FLOX regimen also being the most frequently described protocol in 34 patients (53.1%). Other chemotherapy regimens were indicated, with CAPOX being the most prevalent in 11 patients (17.2%), followed by FOLFOX in 3 patients (4.7%), Irinotecan in 3 patients (4.7%), and other regimens in 6 patients (9.4%) (Table 3).

Table 3.

Neoadjuvant chemotherapy, adjuvant chemotherapy, recurrence sites, percentage of recurrence after hepatectomy at 1, 3, and 5 years

	n (%)
Total	64 (100.0%)
Neoadjuvant chemotherapy
No	21 (32.8%)
Yes	43 (67.2%)
FLOX	32 (50.0%)
Others	11 (17.2%)
Adjuvant chemotherapy
No	7 (10.9%)
Yes	57 (89.1%)
FLOX	34 (53.1%)
CAPOX	11 (17.2%)
FOLFOX	3 (4.7%)
Irinotecan	3 (4.7%)
Others	6 (9.4%)
Relapse site	35 (54.7%)
Hepatic	28 (80.0%)
Pulmonary	15 (42.9%)
Lymph node	11 (31.4%)
Others	4 (11.4%)
Years after hepatectomy: recurrence
Relapse within 1 year	22 (34.4%)
Relapse within 3 years	33 (51.6%)
Relapse within 5 years	35 (54.7%)

Source: Prepared by the author (2024) with epidemiological data expressed as absolute and percentage frequencies

Of the total sample, 35 patients (54.7%) experienced recurrence after hepatectomy for metastatic colorectal cancer, with the liver being the main site of involvement (n = 28, 80.0%), followed by the lung, with 15 patients (42.9%) and lymph node with 11 patients (31.4%) (Table 3).

The median disease-free survival was only 16 (95% CI = 11–22) months, with 45.2% of patients being disease-free at a median follow-up time of 27.7 ± 4.2 (95% CI = 19.4–36.0) months.

Regarding the expression of the markers, ROC curves did not show significant cutoffs for HSPA9 (p = 0.842), FUS (p = 0.309), p53 (p = 0.377) and pS6 (p = 0.154) based on the results of the analysis of the expression of the biomolecular markers analyzed, with the outcome being the presence of disease-free survival (Table 4).

Table 4.

Results of the expression of biomolecular markers pS6, BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS and HSPA9 in patients undergoing hepatectomy for metastatic colorectal cancer

	Total
HSPA9 (< 300 vs. 300)
Negative	17 (27.9%)
Positive	44 (72.1%)
FUS (Up to 70 vs. > 70)
Negative	28 (47.5%)
Positive	31 (52.5%)
P53 (Up to 30 vs. > 30)
Negative	32 (60.4%)
Positive	21 (39.6%)
CERBB2 (0 vs. 1–3)
Negative	49 (96.1%)
Positive	2 (3.9%)
PMS2
Retention	57 (98.3%)
Loss	1 (1.7%)
MSH6
Retention	54 (100.0%)
Loss	0 (0.0%)
pS6 (0 vs. > 0)
Negative	46 (79.3%)
Positive	12 (20.7%)
BRAF V600E
Negative	44 (78.6%)
Positive	12 (21.4%)

Source: Prepared by the author (2024). Data expressed as absolute and percentage frequencies

Since ROC analysis did not identify statistically significant cutoff points for HSPA9 (p = 0.842), FUS (p = 0.309), p53 (p = 0.377), and pS6 (p = 0.154), alternative thresholds were defined based on established biological and histopathological criteria.

For HSPA9, an H-score of 300 (n = 44, 72.1%) was adopted, representing the upper limit of the scale and maximum protein expression.

For FUS, the median of the 70% cohort (n = 31, 52.5%) was used as a data-driven threshold to dichotomize expression.

For p53, a score of 30 (n = 21, 39.6%) was adopted according to standard immunohistochemical interpretation: scores between 1 and 30 indicate weak and focal expression compatible with non-mutated protein, while scores above 30 indicate overexpression compatible with a mutated pattern.

For pS6, any positive staining above zero was considered positive. This threshold was adopted based on the biological premise that any detectable activation of the PI3K/AKT/mTORC1 pathway — for which pS6 serves as a surrogate marker (Fig. 1) — may carry clinical relevance, given the role of this pathway in apoptosis resistance, cell proliferation, and chemotherapy resistance in mCRC. This approach is further supported by evidence from Hirashita et al. (2021), who demonstrated that pS6 immunostaining-based detection is useful for predicting sensitivity to MEK inhibitors in RAS/BRAF-mutant colorectal cancer, suggesting that even low-level pS6 expression reflects biologically meaningful pathway activation [19]. Notably, all 12 pS6-positive patients in the present cohort had primary tumors located in the left colon, sigmoid, or rectum — a distribution consistent with the established literature on the differential activation of the PI3K/AKT/mTORC1 pathway according to tumor sidedness in CRC, reinforcing the biological plausibility of this threshold. ROC curve analysis did not identify a statistically significant cutoff for pS6 (p = 0.154), which is expected given the small sample size and the low prevalence of pS6 positivity in this cohort. In the absence of a ROC-derived threshold, the adoption of any-positive staining as the cutoff represents an a priori biologically grounded decision rather than a data-driven one, which avoids circular reasoning and is appropriate for an exploratory study of this nature.

For BRAF V600E, strong and diffuse staining was required, according to validated immunohistochemical criteria; cases without this pattern were classified as negative.

It is important to distinguish between the nature of the thresholds used for each marker in this study. For BRAF V600E, the strong and diffuse staining pattern used as a positivity criterion is a morphologically established standard with well-documented diagnostic reproducibility. For MSH6 and PMS2, complete absence of expression in neoplastic cells — used to define MMR loss — is a clinically validated and internationally standardized criterion. For CERBB2, the 0 versus 1–3 + scoring system follows standard HER2 assessment guidelines. In contrast, the thresholds adopted for HSPA9 (H-score ≥ 300), FUS (H-score ≥ 70), p53 (H-score > 30), and pS6 (any positivity above zero) were not derived from statistically significant ROC curves in this cohort and should be explicitly classified as exploratory, biologically motivated thresholds. They were selected based on established literature and biological rationale, not from data-driven optimization in the present dataset, and therefore cannot be considered validated clinical cutoffs. These thresholds require prospective validation in larger, independent cohorts before any clinical application.

The results of the immunohistochemistry reactions were as follows: HSPA9 negative in 17 patients (27.9%) and positive in 44 (72.1%); FUS negative in 28 patients (47.5%) and positive in 31 (52.5%); p53 negative in 32 patients (60.4%) and positive in 21 patients (39.6%), no patients had null expression; CERBB2 was negative in 49 patients (96.1%) and positive in 2 (3.9%); PMS2 showed retention (maintenance of expression) in 57 patients (98.3%) and loss of expression in 1 (1.7%); MSH6 showed retention (maintenance of expression) in 54 patients (100.0%) and no loss of expression; pS6 was negative in 46 patients (79.3%) and positive in 12 (20.7%); BRAF V600E was negative in 44 patients (78.6%) and positive in 12 (21.4%).

Representative photomicrographs of each immunohistochemical marker are illustrated in Fig. 2, with panels A through H displaying the characteristic staining patterns of HSPA9, FUS, CERBB2, p53, PMS2, MSH6, pS6, and BRAF V600E, respectively.

Fig. 2.

Photomicrographs of immunohistochemical markers in hepatectomies for metastases of colonic adenocarcinoma. Immunohistochemistry, 200×. A — HSPA9: diffuse cytoplasmic staining with H-score of 300. B — FUS: diffuse nuclear and cytoplasmic staining (H-score > 70). C — CERBB-2: only 3 + case, with strong and complete membranous staining in more than 10% of cells. D — p53: diffuse and strong nuclear staining (H-score > 30). E — PMS-2: single negative case, with neoplastic cells on the left side showing no PMS-2 immunoexpression and stromal cells on the right side staining positive (cases in which no positivity was observed in either compartment were considered inconclusive). F — MSH-6: diffuse nuclear positivity (no case with loss of MSH-6 expression). G — pS6: weak and focal cytoplasmic staining; pS6 is much less expressed in neoplastic cells, with positivity restricted to a few cases. H — BRAF V600E: strong and diffuse cytoplasmic staining in neoplastic cells for the mutated BRAF V600E protein. Source: Prepared by the author (2024)

Figure 2.

Regarding the outcomes studied in this work, the results were disease-free survival and overall survival, the latter calculated by the difference between the patient’s diagnosis date and the date of the last consultation for those who did not have a recurrence, and the date of recurrence or disease progression for those who presented with disease recurrence.

During the analysis of this study, it was identified that there was no statistical significance between the variables of sex, age, primary tumor site or presence of synchronous lesions at the time of diagnosis and the type of initial approach to the lesions; that is, these variables in the present study were not associated with disease recurrence after hepatectomy.

Similarly to these variables analyzed, at 1 year, 3 years, and 5 years post-hepatectomy, the number of liver lesions (up to 3, from 3 to 5, or greater than 5) at the time of diagnosis, length of hospital stay after hepatectomy (up to 7 days, between 7 and 15 days, and greater than 15 days), whether or not there were postoperative complications, and the postoperative complications in each grade of the Clavien-Dindo Classification also did not have statistical significance.

Regarding the indication for adjuvant chemotherapy in hepatectomies for metastatic colorectal cancer, from January 2012 to December 2022 at the Haroldo Juaçaba Hospital/Ceará Cancer Institute (HJ/ICC), all patients who were able to begin follow-up with clinical oncology after surgery received an indication for adjuvant chemotherapy.

The patients who did not receive chemotherapy were the 5 who died and 1 patient who was prescribed chemotherapy but developed obstructive jaundice, requiring biliary drainage, but died late, without being able to undergo the prescribed adjuvant chemotherapy treatment.

During the review of medical records, problems were identified such as loss to follow-up of patients due to lack of financial resources, abandonment of treatment due to disbelief or lack of understanding, hospitalizations for other reasons, and long-term deaths, including deaths from COVID-19, strokes, and cardiovascular events, suggesting that patients face additional challenges beyond the disease itself, which may affect treatment outcome and survival.

Regarding the molecular markers, pS6, BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS and HSPA9, the statistically significant data was from patients with pS6 positivity, who presented lower disease-free survival at 1 (p = 0.032), 3 (p = 0.038) and 5 years (p = 0.018) (Table 5).

Table 5.

Molecular markers pS6, BRAF V600E, p53, CERBB2, MSH6, PMS2, FUS, and HSPA9 associated with disease-free survival at 1, 3, and 5 years in patients undergoing hepatectomy for metastatic colorectal cancer

	Disease-free survival at 1 year		p-	3-year disease-free survival		p-	5-year disease-free survival		p-
	No	Yes	Value	No	Yes	Value	No	Yes	Value
HSPA9 (< 300 vs. 300)
Negative	12 (30.8%)	5 (22.7%)	0.501	8 (26.7%)	9 (29.0%)	0.837	8 (28.6%)	9 (27.3%)	0.910
Positive	27 (69.2%)	17 (77.3%)		22 (73.3%)	22 (71.0%)		20 (71.4%)	24 (72.7%)
FUS (Up to 70 vs. > 70)
Negative	15 (40.5%)	13 (59.1%)	0.168	12 (42.9%)	16 (51.6%)	0.501	11 (42.3%)	17 (51.5%)	0.482
Positive	22 (59.5%)	9 (40.9%)		16 (57.1%)	15 (48.4%)		15 (57.7%)	16 (48.5%)
P53 (Up to 30 vs. > 30)
Negative	19 (57.6%)	13 (65.0%)	0.592	15 (60.0%)	17 (60.7%)	0.958	16 (65.2%)	17 (56.7%)	0.528
Positive	14 (42.4%)	7 (35.0%)		10 (40.0%)	11 (39.3%)		8 (34.8%)	13 (43.3%)
CERBB2 (0 vs. 1–3)
Negative	30 (93.8%)	19 (100.0%)	0.266	24 (100.0%)	25 (92.6%)	0.174	22 (100.0%)	27 (93.1%)	0.209
Positive	2 (6.3%)	0 (0.0%)		0 (0.0%)	2 (7.4%)		0 (0.0%)	2 (6.9%)
PMS2
Retention	37 (100.0%)	20 (95.2%)	0.181	28 (100.0%)	29 (96.7%)	0.330	26 (100.0%)	31 (96.9%)	0.363
Loss	0 (0.0%)	1 (4.8%)		0 (0.0%)	1 (3.3%)		0 (0.0%)	1 (3.1%)
MSH6
Retention	34 (100.0%)	20 (100.0%)	1.000	25 (100.0%)	29 (100.0%)	1.000	23 (100.0%)	31 (100.0%)	1.000
Loss	0 (0.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)		0 (0.0%)	0 (0.0%)
pS6 (0 vs. > 0)
Negative	27 (71.1%)	19 (95.0%) *	0.032	19 (67.9%)	27 (90.0%) *	0.038	17 (65.4%)	29 (90.6%) *	0.018
Positive	11 (28.9%) *	1 (5.0%)		9 (32.1%) *	3 (10.0%)		9 (34.6%) *	3 (9.4%)
BRAF V600E
Negative	30 (81.1%)	14 (73.7%)	0.523	22 (78.6%)	22 (78.6%)	1.000	20 (76.9%)	24 (80.0%)	0.780
Positive	7 (18.9%)	5 (26.3%)		6 (21.4%)	6 (21.4%)		6 (23.1%)	6 (20.0%)

Source: Prepared by the author (2024). *p < 0.05, Fisher’s exact test or Pearson’s chi-square test (n, %)

In Table 5, regarding the first year, of the patients with positive pS6, 11 (91.7%) relapsed and only 1 (8.3%) did not relapse, compared to pS6-negative patients, of whom 27 (58.7%) relapsed and 19 (41.3%) did not relapse (p = 0.032).

After 3 years, 9 (75.0%) of the pS6-positive patients relapsed versus 19 (41.3%) of the pS6-negative patients (p = 0.038). After 5 years, 9 (75.0%) of the pS6-positive patients relapsed versus 17 (37.0%) of the pS6-negative patients (p = 0.018) (Table 5).

Regarding pS6, a univariate analysis was performed comparing pS6 with other factors of this marker and with clinicopathological characteristics. When comparing the analysis with the characteristics of sex, age, primary tumor, synchronous lesions, and type of approach to the synchronous lesion, no statistically significant data were observed, as well as data on the number of liver lesions, length of hospital stay, and postoperative complications, which also showed no statistical significance.

Following a univariate analysis of the pS6 marker, still in comparison with neoadjuvant or adjuvant chemotherapy and the chemotherapy protocols used in this study, the data were not statistically significant (Table 6).

Table 6.

Univariate analysis: pS6 versus disease-free survival at 1 year, 3 years, and 5 years, plus comparison with markers

	pS6		p-
	Negative	Positive	Value
Disease-free survival at 1 year
No	27 (58.7%)	11 (91.7%) *	0.032
Yes	19 (41.3%) *	1 (8.3%)
3-year disease-free survival
No	19 (41.3%)	9 (75.0%) *	0.038
Yes	27 (58.7%) *	3 (25.0%)
5-year disease-free survival
No	17 (37.0%)	9 (75.0%) *	0.018
Yes	29 (63.0%) *	3 (25.0%)
HSPA9
Negative	13 (29.5%)	3 (25.0%)	0.757
Positive	31 (70.5%)	9 (75.0%)
FUS
Negative	25 (56.8%) *	2 (16.7%)	0.014
Positive	19 (43.2%)	10 (83.3%) *
P53
Negative	24 (58.5%)	8 (72.7%)	0.390
Positive	17 (41.5%)	3 (27.3%)
CERBB2
Negative	38 (95.0%)	10 (100.0%)	0.470
Positive	2 (5.0%)	0 (0.0%)
PMS2
Negative	43 (97.7%)	12 (100.0%)	0.598
Positive	1 (2.3%)	0 (0.0%)
MSH6
Negative	41 (100.0%)	12 (100.0%)	1.000
Positive	0 (0.0%)	0 (0.0%)
BRAF V600E
Negative	37 (84.1%)	7 (58.3%)	0.054
Positive	7 (15.9%)	5 (41.7%)

Source: Prepared by the author (2024). *p < 0.05, Fisher’s exact test or Pearson’s chi-square (n, %)

Regarding the assessment by pS6 expression, overall survival, from the date of surgery and last consultation after the second outcome analysis for each patient, was 23.00 months (95% CI = 18.33–27.66) with a follow-up of 35.09 ± 5.02 months (95% CI = 25.25 ± 44.94) (Figs. 3 and 4). When stratified by pS6 status, median disease-free survival was 13.80 months (95% CI = 8.62–18.98) in the pS6-positive group versus 20.90 months (95% CI = 14.62–27.18) in the pS6-negative group (p = 0.312). In the pS6-positive group, 9 events (75.0%) and 3 censored patients (25.0%) were recorded; in the pS6-negative group, 17 events (37.0%) and 29 censored patients (63.0%). Regarding overall survival stratified by pS6 status, median OS was 14.00 months (95% CI = 8.46–19.54) in the pS6-positive group versus 23.00 months (95% CI = 18.13–27.87) in the pS6-negative group (p = 0.045). In the pS6-positive group, 9 events (75.0%) and 3 censored patients (25.0%) were recorded; in the pS6-negative group, 11 events (23.9%) and 35 censored patients (76.1%). These findings confirm that pS6 positivity is associated with both shorter disease-free survival and shorter overall survival, consistent with the multivariate analysis results.

Fig. 4.

Overall survival of patients who underwent hepatectomies for metastatic colorectal cancer, assessed by pS6 expression (solid line = positive expression; dashed line = negative expression), 1-year disease-free survival (solid line = patients with disease-free survival; dashed line = patients without disease-free survival), and administration of adjuvant chemotherapy (solid line = yes; dashed line = no). Source: Prepared by the author (2024). *p < 0.05, Cox regression. HRa = adjusted hazard ratio

Table 7 shows the results in the form of adjusted hazard ratio for disease-free survival at 1 year, 3 years, and 5 years, analyzing the location of the primary tumor, postoperative complications, adjuvant chemotherapy, and pS6, in which pS6 showed excellent performance and, even when comparing the results running with other predictor variables, it remained significant.

Table 7.

Multivariate analysis, with results in the form of adjusted hazard ratio for disease-free survival at 1 year, 3 years, and 5 years, analyzing the location of the primary tumor, postoperative complications, adjuvant chemotherapy, and pS6

	1 year				3 years				5 years
	p-value	HRa	95% CI		p-value	HRa	95% CI		p-value	HRa	95% CI
Disease-free survival
Location of the primary tumor	0.123	4.522	0.664	30.769	0.234	4.244	0.393	45.834	0.452	2.505	0.229	27.353
Postoperative complications	0.256	2.203	0.563	8.616	0.014	8.671	1.560	48.185	0.029	6.912	1.217	39.257
adjuvant chemotherapy	0.997	1.666	0.167	16.664	0.997	2.290	0.229	22.900	0.997	2.713	0.271	27.130
pS6	0.058	9.169	0.925	90.866	0.028	9.710	1.279	73.716	0.018	11.857	1.530	91.897

Source: Prepared by the author (2024). *p < 0.05, Cox regression. HRa = adjusted hazard ratio

Underlined and italicized values indicate statistically significant results (p < 0.05)

Ultimately, in a multivariate analysis using an adjusted hazard ratio of the risk of death in relation to the number of liver lesions, neoadjuvant chemotherapy, adjuvant chemotherapy, chemotherapy protocol used, 1-year disease-free survival, and pS6 expression, adjuvant chemotherapy reduced the risk of death by 0.052 times, the absence of 1-year disease-free survival increased the risk by 3.641 times, and pS6 positivity increased the risk of death by 4.547 times, independently of the other variables studied. The prognostic predictors were adjuvant chemotherapy, 1-year disease-free survival, and pS6 positivity (Table 8).

Table 8.

Results in the form of adjusted hazard ratio of the risk of death in relation to the number of liver lesions, neoadjuvant chemotherapy, adjuvant chemotherapy, chemotherapy protocol used, 1-year disease-free survival, and pS6 expression (positive)

	p-value	HRa	95% CI
Risk of death
Number of liver lesions	0.698	1.180	0.513	2.715
Neoadjuvant chemotherapy	0.330	1.419	0.701	2.871
Adjuvant chemotherapy	< 0.001	0.052	0.016	0.168
Adjuvant chemotherapy protocol	0.518	0.957	0.836	1.094
Disease-free survival at 1 year	0.001	3.641	1.654	8.016
pS6 expression (positive)	0.001	4.547	1.878	11.010

Source: Prepared by the author (2024). *p < 0.05, Cox regression. HRa = adjusted hazard ratio

Underlined and italicized values indicate statistically significant results (p < 0.05)

Discussion

Advances in molecular oncology have significantly impacted the screening, treatment, and prognosis of CRC [15]. CRC is characterized by well-defined risk factors, slow progression, and pre-malignant lesions treatable by colonoscopy; however, its incidence has increased [3–5].

Many patients present with metastatic disease, predominantly liver involvement, at the time of diagnosis. Hepatectomy remains the gold standard for resectable lesions. For unresectable lesions, when indicated, conversion chemotherapy followed by achievement of resectability demonstrates non-inferiority in terms of disease-free survival and overall survival compared with initial resection [16, 17].

Treatment strategies for mCRC depend on the clinical presentation and stage of the patient. Potentially, starting with the primary tumor site, followed by treatment of liver lesions, using a “liver-first” approach or a concurrent approach [22].

In this study, the most prevalent approach was resection of the primary tumor, followed by hepatectomy (n = 27, 42.2%) and initial hepatic approach (n = 11, 17.2%). Many patients underwent surgery for obstructive lesions or presented with obstructive tumors at the initial consultation, highlighting the need for early diagnosis [5]. After hepatectomy for metastatic colon neoplasia, the liver remained the main site of recurrence, followed by the lungs, lymph nodes, and other sites.

The median overall survival in Fig. 3 was 23 months, with a mean follow-up of 35 months, consistent with the literature. Takamizawa et al. [23] reported that, of 342 patients in the primary resection group and 36 in the conversion group, both experienced hepatic recurrences, with 139 patients (41%) in the primary resection group and 17 (47%) in the conversion group experiencing hepatic recurrence.

Fig. 3.

Overall survival of disease-free patients after hepatectomy for metastatic colorectal cancer versus months. Source: Prepared by the author (2024). The median overall survival was 23.00 (95% CI = 18.33-27.66) months with a median follow-up time of 35.09±5.02 (95% CI = 25.25±44.94) months

Using a cutoff point of 300, 17 (27.9%) patients had a negative result for HSPA9 and 44 (72.1%) had a positive result for HSPA9. Esfahanian et al. [13] described the role of HSPA9 in maintaining protein synthesis and cellular homeostasis in response to environmental stress, thus protecting cancer cells and influencing the biology and prognosis of early and advanced stage prostate, breast, and colorectal cancers. One of its oncogenic mechanisms is its ability to bind to cytoplasmic p53, inhibiting its tumor suppressor function in the nucleus.

With a median cutoff point of 70, 28 (47.5%) patients had a negative result for FUS and 31 (52.5%) had a positive result for FUS. Sato et al. [10] demonstrated a proto-oncogenic interaction between β- catenin and FUS, gradually decreasing the depth of the crypts with increasing expression of the FUS protein. The expression pattern of FUS in adenomas reflected that of accumulated β -catenin. FUS was expressed in the nuclei of colorectal adenocarcinoma cells from patients with familial adenomatous polyposis and sporadic CRC.

Univariate analysis revealed a significant association between FUS positivity and pS6 positivity (p = 0.014), suggesting a possible interaction between the FUS/β-catenin-mediated proto-oncogenic pathway and the PI3K/AKT/mTORC1 pathway, for which pS6 serves as a surrogate marker (Fig. 1). This observation is biologically plausible in light of existing evidence on the crosstalk between these two pathways in CRC. The Wnt/β-catenin and PI3K/AKT/mTORC1 pathways are both critically involved in CRC development and are connected at multiple levels through shared upstream and downstream effectors. Their reciprocal regulation represents one of the main mechanisms of resistance to selective inhibitors in CRC, and in settings driven by Wnt/β-catenin genetic alterations, the relationship between these two pathways has been described as so close that some authors have proposed they should be considered a unique therapeutic target [24]. In this context, FUS — which interacts with β-catenin and modulates its proto-oncogenic activity in CRC [10] — may plausibly contribute to co-activation of the PI3K/AKT/mTORC1 pathway, reflected by elevated pS6 levels. However, it is important to emphasize that the association observed here is purely statistical and derived from a small retrospective cohort. The biological mechanism linking FUS and pS6 in this clinical setting remains speculative and has not been directly demonstrated. Mechanistic studies and validation in larger prospective cohorts are needed before any conclusion about pathway convergence can be drawn.

Using a cutoff point of 30, 32 patients (60.4%) had a negative result and 21 (39.6%) a positive result for p53, with no null expression observed. Vogelstein’s work on the APC-KRAS-DCC-P53 mutation sequence [6] was fundamental to understanding colorectal carcinogenesis. Ottaiano et al. [8] summarized that p53 is one of the most frequently mutated genes in CRC, leading to loss or gain of function; however, a meta-analysis of 140 studies (2013–2022) showed that p53 has no prognostic value in metastatic CRC.

Any positive pS6 staining was considered positive; 46 (79.3%) patients had a negative result and 12 (20.7%) had a positive result, all located in the left colon (n = 3), sigmoid colon (n = 3) and rectum (n = 6).

The BRAF V600E mutation was classified as positive due to strong and diffuse staining (otherwise negative); 44 (78.6%) patients had a negative result and 12 (21.4%) a positive result. Primary tumors were located in the left colon (n = 1), sigmoid colon (n = 6), and rectum (n = 5), with a relatively high positivity rate. BRAF activation increases S6 phosphorylation, leading to elevated pS6 levels. Hirashita et al. [19], using CRC organoids, explored alterations in pS6 as a biomarker for MEK inhibition, aiming to identify patients highly sensitive to MEK inhibitors and thus improve the therapeutic benefit for CRC patients with RAS/BRAF mutation. This highlights the need for biomarkers to predict the efficacy of MEK inhibitors. The study suggested that immunohistochemical detection of pS6 could predict sensitivity to trametinib, a MEK inhibitor used in the treatment of BRAF V600E-mutant metastatic melanoma.

Regarding CERBB2, the analysis of scores of zero versus 1–3 showed 49 (96.1%) negative results and only 2 (3.9%) positive, with one primary tumor in the rectum and the other in the transverse colon. This points to a limitation of the study due to the small number of patients; however, the literature supports the role of CERBB2 as a marker, particularly in mCRC. Kaur et al. [14] described an observational study of 50 CRC cases analyzing HER2 expression for diagnostic utility. Positive HER2 expression was observed in 32%, most frequently in rectal tumors. The prognostic role of HER2 in mCRC requires further investigation; however, strategies to improve survival and reduce toxicity are being developed [20].

PMS2 showed retained expression in 57 (98.3%) patients and loss in only 1 (1.7%) patient, with the primary tumor located in the right colon. All 54 patients (100%) showed retained expression of MSH6. Kanthan et al. [15] described how microsatellite instability (MSI) affects genes involved in proliferation, cell cycle, apoptosis, and DNA repair. Inactivation of repair genes can occur through hereditary or somatic mutations. Lynch syndrome, the most common hereditary cause of CRC (representing 3% of cases), is associated with mutations in the MLH1, MSH2, MSH6, and PMS2 genes. The BRAF test is indicated because this gene rarely shows mutations in Lynch syndrome.

This study highlighted the prognostic value of pS6. Median overall survival was 23 months (median follow-up of 35 months), suggesting high persistent morbidity and mortality in CRC even after treatment. pS6 was associated with one-year disease-free survival and overall survival, suggesting its potential as a valuable prognostic biomarker. Wiesweg et al. [25] analyzed 160 patients with advanced CRC prospectively enrolled in a biomarker program and found that high phosphorylation of p70 S6 β -1 kinase was an independent predictor of poor prognosis for right-sided primary tumors, with reduced overall survival [25].

From a biological standpoint, pS6 positivity reflects the activation of the PI3K/AKT/mTORC1 pathway (Fig. 1), which has been associated with resistance to apoptosis, increased cell proliferative capacity, and resistance to chemotherapy. This biological context is consistent with the worse disease-free survival and higher mortality risk observed among pS6-positive patients in the present study, and may provide a plausible mechanistic basis for these associations. It is important to note, however, that pS6 likely reflects an aggressive tumor biology rather than directly driving recurrence or mortality — its prognostic value resides in its role as a surrogate marker of pathway activation, not as a proven causal mediator of clinical outcomes. However, these findings must be interpreted with caution. Only 12 patients (20.7%) had pS6-positive tumors, and this small subgroup substantially limits the statistical stability of the Cox regression estimates. The wide confidence intervals observed — particularly for the hazard ratios at 3 years (HRa = 9.710; 95% CI = 1.279–73.716) and 5 years (HRa = 11.857; 95% CI = 1.530–91.897) for disease-free survival, and for overall survival (HRa = 4.547; 95% CI = 1.878–11.010) — reflect this instability and should be explicitly acknowledged. Although the associations reached statistical significance, the magnitude of the point estimates and the breadth of their confidence intervals indicate that these results are hypothesis-generating and require prospective validation in larger cohorts before any clinical inference can be drawn.

This study evaluates pS6 as a candidate biomarker in isolation. It is important to recognize, however, that the field of biomarker research in oncology has evolved beyond isolated molecular variables towards more integrated and multimodal predictive frameworks. A recent study developed a multimodal digital biopsy model integrating primary tumor radiomics with clinical factors for non-invasive preoperative prediction of occult peritoneal metastases in locally advanced gastric cancer, demonstrating robust performance in multiple validation cohorts [26]. Although this work addresses a distinct tumor type and clinical issue, it is methodologically relevant in illustrating how biologically informed models that combine multiple data sources can refine risk assessment beyond isolated conventional staging, suggesting that the prognostic value of pS6 identified here could be enhanced in future integrated approaches.

Along the same lines, a multimodal virtual biopsy model based on artificial intelligence, integrating radiomic and clinical data, was developed for risk stratification in gastric cancer, demonstrating superior predictive accuracy compared to single-modality models [27]. These examples highlight a broader trend in translational oncology research towards composite and integrated prediction strategies, suggesting that the prognostic value of pS6 identified in the present study could be further enhanced in future studies that incorporate it into multimodal predictive frameworks applied to liver metastases from colorectal cancer.

The recurrence findings reported here are potentially important, given the emphasis on disease-free survival after hepatectomy. In this context, recent literature on recurrence prediction in gastrointestinal cancers offers relevant methodological references. A radiomics-based ensemble model incorporating clinical and imaging data has demonstrated robust performance for predicting early postoperative recurrence in locally advanced gastric cancer in multiple validation cohorts [28]. Similarly, an interpretable multimodal fusion model integrating clinical, radiomic, and pathomic data significantly outperformed single-modality approaches in predicting early recurrence, while also providing biological insights through transcriptomic analysis [29]. These studies, although focused on gastric cancer, are methodologically relevant in highlighting the growing consensus that recurrence prediction models should combine predictive performance with biological interpretability—a principle that can guide future studies incorporating pS6 into integrated prediction frameworks after hepatectomy for metastatic colorectal cancer.

In this sense, a deep learning-based digital biopsy model, integrating histopathological features with clinical variables, demonstrated the utility of clinically actionable and biologically informed postoperative risk stratification tools for surveillance after surgery [30]. Collectively, these advances suggest that pS6 may represent a promising component of future integrated recurrence prediction frameworks for patients undergoing hepatectomy for metastatic colorectal cancer, rather than serving exclusively as an isolated marker.

The limitations of this study include the small sample size (n = 64), which restricts the statistical power of multivariate analyses and raises concerns about model stability, sparse data bias, and the reliability of hazard ratio estimates. Of particular concern is the small number of pS6-positive patients (n = 12, 20.7%), which substantially limits the statistical stability of the Cox regression models in which pS6 was included. The resulting wide confidence intervals — spanning more than two orders of magnitude for disease-free survival at 3 and 5 years — should be explicitly acknowledged as a major source of uncertainty, and the hazard ratio point estimates should not be interpreted in isolation. The multivariate models include several variables relative to the number of events, which increases the risk of overfitting; although variables were selected a priori based on clinical relevance, the risk of model instability cannot be excluded in a cohort of this size. The proportional hazards assumption was not formally tested, which represents an additional methodological limitation. These findings should therefore be interpreted as hypothesis-generating, not definitive. The retrospective design and data collection from electronic medical records may affect data accuracy. Loss to follow-up due to treatment abandonment, financial constraints, and deaths from non-oncological causes—including COVID-19, stroke, and cardiovascular events—may have introduced informational censoring and affected survival estimates. As this is a single-center study conducted at a reference institution in Northeast Brazil, the generalizability of the results to other populations and healthcare settings may be limited. Additionally, the absence of a comprehensive molecular profile, including RAS mutation status, microsatellite instability, and tumor mutational burden, in all patients limited a more complete biological characterization of the context in which pS6 expression exerts its prognostic effect. Furthermore, several clinicopathological variables with established prognostic relevance — including surgical margin status, size and bilobar distribution of liver metastases, preoperative CEA levels, and RAS mutation status — were not collected or could not be reliably assessed in this cohort, limiting a more comprehensive prognostic analysis. RAS mutational status could not be evaluated by immunohistochemistry due to unsatisfactory staining results, and will be addressed in future prospective work. Immunohistochemical evaluation was performed by a single pathologist blinded to clinical data; the absence of a second independent reader means that interobserver agreement could not be assessed, which is an acknowledged limitation of the present study. Prospective, multicenter studies with larger samples, incorporating standardized molecular profiling, are needed to externally validate the prognostic role of pS6 in liver metastases from colorectal cancer.

Conclusion

In this cohort of patients undergoing hepatectomy for metastatic colorectal cancer, median overall survival was 23 months, with a median follow-up of 35 months, reflecting the high and persistent mortality burden of this condition even after surgical treatment with curative intent. Among the eight molecular markers evaluated by immunohistochemistry, pS6 was the only one independently associated with clinical outcomes: pS6 positivity was associated with worse disease-free survival at 1, 3, and 5 years, and was an independent predictor of both recurrence and mortality in multivariate analysis. These associations are consistent with the biological role of pS6 as a surrogate marker of PI3K/AKT/mTORC1 pathway activation, suggesting that this pathway may contribute to a more aggressive tumor phenotype in this setting, though a direct causal relationship cannot be established from the present data.

A statistically significant association between pS6 and FUS positivity was also identified, suggesting a possible convergence of oncogenic pathways that may contribute to a more aggressive tumor phenotype and warrants further investigation. Postoperative complications were independent predictors of worse disease-free survival, underscoring the prognostic importance of surgical quality and perioperative morbidity control. Adjuvant chemotherapy was the strongest independent predictor of improved overall survival, confirming its essential role in the multimodal management of mCRC after hepatectomy.

Given the retrospective design, single-center setting, and limited sample size, all findings should be interpreted as hypothesis-generating. Prospective, multicenter studies with larger samples and standardized molecular profiling are needed for external validation of pS6 as a prognostic biomarker before its clinical application in patients undergoing hepatectomy for metastatic colorectal cancer.

Authors’ contributions

R.H.C.T.M. conceived and designed the study, collected and analyzed the data, interpreted the results, and drafted the manuscript. P.G.B.S. performed the statistical analysis. H.F.C.N., R.C.S.N., J.R.M.T.M., M.J.B.B., and C.A.D. contributed to data collection and critical revision of the manuscript. C.G.H. supervised the study, contributed to the interpretation of results, and critically revised the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funding was received for the research, authorship, and/or publication of this article.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Hospital Haroldo Juaçaba / Instituto do Câncer do Ceará (CAAE: 67319223.0.0000.5528; Ethics Committee Opinion No. 6.989.739, approved on August 7, 2024). All procedures adhered to Brazilian ethical guidelines for research involving human beings (Resolution CNS 466/12) and to the principles of the Declaration of Helsinki and the Nuremberg Code.

Competing interests

The authors declare no competing interests.