Association of peripheral blood cell count-derived ratios, biomarkers of inflammatory response and tumor growth with outcome in previously treated metastatic colorectal carcinoma patients receiving cetuximab

Introduction

Colorectal carcinoma represents one of the most common malignant tumors [1]. Despite the progress in screening and early diagnosis, many patients are still diagnosed with advanced disease. Moreover, even among patients initially presenting with early disease many eventually experience relapse in spite of surgical therapy, adjuvant chemotherapy or, in the case of rectal carcinoma, neoadjuvant radiotherapy. Thus, the mortality for colorectal carcinoma remains high. On the other hand, the outcome of patients with advanced (mostly metastatic) colorectal carcinoma has improved over the past two decades with the advent of active cytotoxic agents and targeted drugs.

The identification of predictive and prognostic biomarkers is a crucial component of the current strategy in the management of cancer patients [2]. With the increasing diversity of treatment options, biomarkers also play an important role in the management of colorectal carcinoma. Molecular biomarkers in the tumor tissue, e.g. RAS mutations, are helping to identify patients likely to benefit from different regimens of available agents [3]. Circulating biomarkers that reflect tumor growth and biological behavior are also important in cancer patients [2].

There is a cumulative body of evidence indicating that, along with molecular biomarkers of cancer cells, biomarkers that reflect the host response to neoplasia are of equal importance. The infiltration of the tumor by lymphocytes, so-called tumor infiltrating lymphocytes (TIL), represents a biomarker predicting outcome across a range of neoplastic disorders [4], [5], [6]. Utilization of TIL is obviously limited by the need to obtain a biopsy. However, it has been demonstrated during the past few years that relative lymphocyte numbers in the peripheral blood expressed as a ratio to other blood elements are also a powerful biomarker predictive of patient prognosis, again across a range of tumor types. Neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), platelet-to-lymphocyte ratio (PLR) and systemic inflammatory index (SII) have been shown to predict outcome in different solid tumors, including colorectal carcinoma [7], [8], [9], [10], [11], [12], [13], [14], [15].

The aim of the present study was to analyze the association of outcome with peripheral-blood cell count (PBC)-derived ratios, other biomarkers of inflammatory response including C-reactive protein (CRP) and urinary neopterin, and biomarkers of tumor growth in a cohort of patients presenting for the treatment with cetuximab after the failure of prior lines of therapy.

Patients and methods

Fifty-one previously treated patients with metastatic colorectal carcinoma, 33 males and 18 females, aged [mean±standard deviation (SD)] 60±11 (range 31–78) years, were treated with cetuximab. between June 2004 and October 2008 in the Department of Oncology and Radiotherapy, Charles University Medical School and Teaching Hospital, Hradec Králové, Czech Republic. All patients had advanced (inoperable) or metastatic colorectal carcinoma that failed at least one line of prior chemotherapy. The utilization of cetuximab at that time was restricted, and drug reimbursement required special approval. One patient was treated with cetuximab monotherapy, and the other 50 patients received the drug in combination with other cytotoxic agents. With the exception of one patient in whom irinotecan could not be administered because of hyperbilirubinemia and who was treated with cetuximab plus 5-fluorouracil and folinic acid, all other 49 patients were administered cetuximab in combination with irinotecan, 5-fluorouracil and folinic acid.

The regimen used in most patients consisted of cetuximab (initial doses of 400 mg/m² and subsequent doses of 250 mg/m² weekly) and biweekly irinotecan (180 mg/m²), folinic acid (200 mg/m²) and 5-fluorouracil (400 mg/m² bolus and 1200 mg/m² in 46-h infusion). The changes and prognostic significance of urinary neopterin, serum magnesium, serum retinol, α-tocopherol and intestinal permeability, or complications of therapy in this cohort or its parts with a shorter follow-up have been reported earlier [16], [17], [18], [19], [20]. PBC with a manual differential count was determined as described earlier [21]. Carcinoembryonic antigen (CEA) was determined by radioimmunoassay as described [17]. Urinary neopterin and creatinine were measured with high-performance liquid chromatography with the result expressed as neopterin/creatinine ratio [21]. Serum calcium, magnesium, bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP), albumin, CRP and lactate dehydrogenase (LDH) were determined using commercial kits on a MODULAR analyzer (Hoffmann-La Roche, Basel, Switzerland). This investigation was part of a project approved by the institutional ethical committee and the patients signed informed consent.

The differences between subgroups of patients were investigated using the Mann-Whitney U test. The correlations were studied with Spearman’s rank correlation coefficient (r_s). The decision on statistical significance was based on p=0.05 level. Survival analysis was performed using Cox regression. Variables with a p-value lower than 0.1 in the univariate analysis were selected for subsequent multivariate analysis. The results were expressed as hazard ratio (HR) and 95% confidence intervals (CI) in both univariate and multivariate analyses. The decision on statistical significance was based on p=0.05 level. The analyses were performed using SAS/STAT^® 9.3 software (SAS Institute Inc., Cary, NC, USA).

Results

Ten patients were treated in the second-line of therapy (i.e. had one prior line of treatment for advanced/metastatic disease), 21 patients in the third line, 13 patients in the fourth line and seven in the fifth line of therapy. The median number of prior lines was 2 (range 1–4), and the median time from the diagnosis of advanced/metastatic disease to the cetuximab treatment start was 22 (range 8–74) months. Forty-seven patients (92%) had liver metastases. A sample of peripheral blood and urine was taken as part of routine care from all 51 patients, 1±1 (range 0–7) days before the first dose, in most cases on the same day, immediately before the administration of premedication. Among the 51 patients in the present cohort, PBC-derived ratios based on manual differential cell count were available immediately before the start of cetuximab in 49 subjects. The hemoglobin concentration was available in 50 patients while urinary neopterin, serum albumin, CRP and CEA concentrations were available in 46, 49, 50 and 50 subjects, respectively.

Table 1 shows the differences in the investigated parameters in 31 patients treated in the second or third line (i.e. with only one line of prior therapy for advanced/metastatic disease) and 20 patients treated in the fourth or higher line of therapy (i.e. with three or more lines of prior therapy). The patients treated in the second or third line of therapy had significantly shorter duration of metastatic disease, higher platelet count, CEA concentration, AST and ALP activity compared to patients treated in the fourth or fifth line.

The CEA concentration before the start of therapy and at least one subsequent measurement during the treatment were available in 46 patients. The median CEA change was −75.3% (range −99.9 to +323.9%). From these 46 patients, 44 patients had initial CEA concentrations above 5 μg/L. Among these 44 patients, CEA decreased by median 82.7% (range 2.8–99.9%) in 40 patients and increased by median 14.9% (range 12.5–323.9%) in four patients. An initial increase followed by a decrease was observed in 10 patients. From these, six with initial CEA higher than 5 μg/L and an increase of more than 10% were considered as surge. The median increase of CEA concentration of 22.6% (range 10.6–56.8%) was observed after a median seven (range 6–14) days.

PBC-derived ratios exhibited mutual correlations as well as correlation with absolute lymphocyte count (Table 2). Serum CRP exhibited significant correlation with urinary neopterin. CRP, but not urinary neopterin, correlated significantly with all four PBC-derived ratios. CRP also correlated positively with nadir CEA concentration and inversely with calcium concentration. Both CRP and urinary neopterin correlated positively with serum CEA and bilirubin concentrations, AST, GGT, ALP and LDH activity and inversely with albumin concentration. Serum magnesium exhibited a negative correlation with absolute lymphocyte counts and a positive correlation with NLR, PLR and SII.

The median duration of cetuximab therapy was 5.1 (range 0.2–17.2) months. Three patients had subsequent liver resection with curative intent. By the time of this analysis, all patients have died. The median survival was 12.1 (range 0.3–110.5) months. The correlations between PBC-derived ratios, other biomarkers of inflammatory response, CEA and survival are shown in Tables 2 and 3.

As shown in Table 3, duration of cetuximab therapy and overall survival were significantly inversely correlated with the nadir CEA concentration during the treatment and positively with the change of CEA during therapy and serum calcium concentration. The overall survival exhibited a significant positive correlation with hemoglobin and albumin concentration and a negative correlation with CRP concentration, GGT and ALP activity.

In univariate analysis, leukocytes and platelet counts, hemoglobin concentration, NLR, urinary hemoglobin, serum calcium, magnesium and bilirubin were significantly associated with survival (Table 4). In the multivariate analysis, only hemoglobin, CEA change and serum bilirubin were independent predictors of survival (Table 5).

Discussion

In the present cohort of limited size, overall survival was correlated with the pretreatment hemoglobin, calcium and albumin concentration, decrease of CEA and inversely with CRP concentration, GGT and ALP activity. However, only hemoglobin, a change of CEA and bilirubin were significant predictors of survival in the multivariate analysis.

CEA was the principal biomarker of tumor mass investigated in the present cohort. The decrease of CEA concentration was a major predictor of prognosis, in agreement with a recent report that investigated patients treated with cetuximab in the first line [22]. The present study also extends the observation of prognostic significance of decreased CEA concentrations after cetuximab to the second or higher line of treatment. The interpretation of changes of CEA concentration is, however, more complex because of the presence of the CEA surge [23]. LDH could not be analyzed in the entire cohort because a different assay was introduced during the time the patients were treated, and patients had LDH values obtained by two non-comparable tests. The survival data of the present cohort of patients were mature and along with the more common Cox regression and Spearman’s correlation of the parameters of survival with laboratory biomarkers could be performed. Although in parameters of borderline statistical significance the results were apparently discordant, the association with survival was, in general, consistent.

Liver is the most common site of metastatic disease in colorectal carcinoma, and liver enzymes including ALT, AST, GGT and ALP as well as serum bilirubin that may indicate liver dysfunction or damage and, indirectly, tumor growth were also analyzed. Neopterin is excreted into the bile [24], and increased neopterin concentrations could be thought to reflect the presence of liver dysfunction as suggested by the correlation between urinary neopterin, bilirubin and the activities of liver enzymes associated with liver dysfunction. It has been demonstrated that increased neopterin concentrations accompany toxic liver injury both in experimental animals [25] and in the clinic [26]. Increased neopterin concentrations have also been associated with viral hepatitis [27]. However, the correlation of these parameters with CRP was even stronger suggesting that the neoplastic infiltration of the liver that reflects high tumor burden accompanied by systemic inflammatory response is the mechanism explaining this association. In fact, both urinary neopterin and CRP correlated with baseline CEA concentration, a parameter that, despite some limitations, approximates the tumor mass.

The findings of the present study may have obvious implications for the management of patients presenting for second or higher line of therapy although the algorithm of the treatment of metastatic colorectal carcinoma has changed, and with anti-EGFR therapy moving to the front-line [22] other treatments, including combination of chemotherapy with aflibercept or bevacizumab, regorafenib or different chemotherapy regimens are being used and the search for new active agents continues [28], [29], [30]. In this population of, sometimes heavily, pretreated patients, the potential benefit of the therapy should be carefully weighted against its complications. The data from the laboratory is essential for the management of cancer patients.

Advanced tumors, including metastatic colorectal carcinoma, are accompanied by high concentrations of biomarkers of inflammatory response that predict poor prognosis [31]. Increased concentrations of biomarkers of inflammatory response are associated with decreased numbers and function of T-lymphocytes in the peripheral blood [32], [33], similarly to the situation in the tumor microenvironment [34], [35]. Notably, increased production of neopterin in vitro and in vivo has been associated with the activity of indoleamine 2,3-dioxygenase (IDO), an enzyme that converts tryptophan to kynurenine [36], [37]. In general, although in some experimental models IDO activation has resulted in tumor cell inhibition [38], and high concentrations of kynurenine are toxic to the cancer cells [39], the predominant effect of IDO activation is now thought to be the suppression of the immune response [40]. In patients with colorectal cancer liver metastases, the serum kynurenine/tryptophan ratio and neopterin concentrations were shown to be increased compared to controls, and the tryptophan concentrations correlated inversely with serum neopterin [36]. In another study reporting a correlation between neopterin concentrations and kynurenine/tryptophan ratio, IDO expression in tumor tissue was associated with parameters of aggressive tumor biology such as lymphatic invasion and presence of lymph node metastases [37]. However, the increase of neopterin concentrations is a non-specific biomarker of immune activation, and along with advanced cancer, high neopterin concentrations were reported across a spectrum of different disorders or conditions including, for example, atherosclerosis and its complications, high age, organ rejection after transplantation, viral infections or autoimmunity [41], [42], [43], [44], [45].

The present data indicate that an inflammatory response is associated with progressing tumor growth in patients with previously treated advanced colorectal carcinoma. The host response is a crucial factor determining the outcome of malignant disease, and both escape from the tumor response and tumor-promoting inflammation have been proposed as factors necessary for tumor progression [46]. Although laboratory medicine plays an important role in the management of cancer patients [2], the assessment of the host antitumor response is inherently difficult. The immune response in the tumor microenvironment that is decisive for the outcome of the host-tumor interaction may be assessed by counting TIL in tumor biopsies or analysis of the presence of inflammatory cells or other biomarkers of inflammatory response in tumor proximal fluids like ascites [35], [47], [48]. For obvious reasons, it is difficult to obtain samples from the tumor microenvironment, and biomarkers of immune and inflammatory response are mostly assessed systemically in the circulation, or, in the case of neopterin, in urine.

Surprisingly, the literature on the correlation between PBC-derived ratios and other biomarkers of inflammatory response is very limited. In agreement with prior reports, PBC-derived ratios have been shown to correlate with CRP, but not with neopterin. Interestingly, neopterin, CRP and PBC-derived ratios are increased across an identical range of disorders spanning from atherosclerosis and its complications to malignant disorders [41], [42]. In earlier studies, we have failed to demonstrate a correlation between urinary neopterin and PBC-derived ratios in several cohorts of patients with breast cancer [21]. However, most of the subjects in this prior study had either early breast cancer or were patients with a history of breast cancer, without active disease. We hypothesized that a correlation between PBC-derived ratios and other biomarkers of inflammatory response might be stronger in patients with advanced disease and high tumor burden. Metastatic colorectal cancer after failure of prior line(s) of therapy represents a model of such disease.

As stated above, similar to prior studies in patients with breast cancer, no correlation was observed between urinary neopterin concentration and any of the PBC-derived ratios investigated, in contrast to the correlations observed with CRP. As mentioned above, an overwhelming majority of patients in the present series had liver metastases, and both urinary neopterin and serum CRP significantly correlated with biomarkers of liver dysfunction including serum bilirubin and albumin concentrations, AST, GGT and ALP activity. In addition, both neopterin and CRP correlated with laboratory parameters that reflect tumor mass, including CEA concentration and LDH activity.

The value of PBC in monitoring hematologic toxicity is obvious, but there is also a wealth of evidence indicating the PBC may also predict patient prognosis. PBC-derived ratios like NLR, LMR, PLR or SII are, in fact, relative lymphocyte counts. The presence of lymphocytes in the tumor tissue (TIL) has been associated with patient prognosis or response to therapy across a range of tumors [49], including colorectal carcinoma [6]. A low number of circulating lymphocytes means reduced capacity of immune response against the tumor. Moreover, neutrophil counts reflect the pro-inflammatory response that is associated with angiogenesis. Depending on a number of factors, the immune and inflammatory responses elicited by the tumor can both inhibit and promote tumor growth [46], and PBC-derived ratios like NLR reflect this balance. Thus, laboratory parameters of lymphocyte presence in the periphery could represent important prognostic and predictive biomarkers that could be used in guiding therapeutic decisions. The prognostic significance of PBC-derived ratios has been amply documented in both early and advanced colorectal cancer [13], [14], [15]. In the present study, NLR was a significant predictor of prognosis in univariate analysis, but in multivariate analysis, hemoglobin was the only hematological parameter independently predicting prognosis, possibly because of the limited size of the cohort. This is in contrast to some other studies [13], [14], [15]. These apparently conflicting results could be related to the size of the cohort, setting (early versus advanced metastatic disease), selection of the end-point (overall survival versus recurrence-free survival) and also the selection of the parameters investigated. For example, in another study in patients with early colorectal cancer the prognostic nutritional index and carbohydrate antigen 19-9 concentration, but none of the PBC-derived ratios were significant predictors of prognosis [50]. Aside from the measurement of PBC, the role of laboratory medicine in predicting or monitoring other important side effects of systemic treatment like gastrointestinal toxicity is more limited [51], [52], [53]. The biomarkers of inflammatory response may predict both the prognosis and complications of therapy [51], [54], but more research should be dedicated to this topic.

The present retrospective study has several limitations. The cohort was rather heterogeneous and because of a selection bias, the patients treated in the second or third line seemed to have a more aggressive disease. This may be a selection phenomenon as patients with aggressive disease did not survive in a good condition long enough to be eligible for the fourth or fifth line of treatment. The size of the cohort was relatively small. The patients were treated before the predictive significance of RAS mutation was discovered [3], and the RAS status is unknown. The absence of information on RAS status might have obscured an association between the biomarkers investigated and the outcome. The treatment of patients was also not homogeneous, and the survival might have been affected by subsequent line(s) of systemic therapy that has undergone substantial transformation and progress during the years when the patients were treated. On the other hand, the survival data are mature, and because no censoring was necessary, unlike in earlier reports in part of the same cohort of patients, a simple correlation analysis between the biomarkers of inflammatory response and survival was possible for screening the prognostic biomarkers of interest. These findings may have implications for the management of patients presenting for second or higher line of therapy, including heavily pretreated patients.

Targeted therapies, including cetuximab, are obviously not free of side effects including skin toxicity, diarrhea, or biochemical abnormalities such as hypomagnesemia and associated hypokalemia and hypocalcemia [18], [55]. Cetuximab is administered mostly in combination with cytotoxic agents like 5-fluorouracil or irinotecan that potentiate diarrhea and lead to bone marrow toxicity. In the present study, only baseline parameters were analyzed, and, therefore, an association of toxicity of cetuximab with the inflammatory response could not be assessed.

In conclusion, the outcome of therapy in the advanced line setting was associated with a decrease of CEA concentration and initial presence of liver dysfunction. The results of the present retrospective analysis also indicate that both urinary neopterin and serum CRP concentrations are correlated with laboratory parameters of liver dysfunction in patients with metastatic colorectal carcinoma and predominantly liver metastases. CRP, but not urinary neopterin correlated with PBC-derived ratios.