Associations between Plasma Insulin-Like Growth Factor Proteins and C-Peptide and Quality of Life in Patients with Metastatic Colorectal Cancer

Colorectal cancer is the fourth most common cancer in the United States and the second leading cause of cancer-related deaths (1). Forty percent of patients will develop metastatic disease during the course of their disease, which generally is incurable. In these patients, treatment should be aimed at prolonging survival while maintaining and improving quality of life (QOL; ref. 2). Although the ability to directly measure QOL has improved (3), few studies have identified predictors of quality among patients with metastatic cancer (4).

Hyperinsulenima and the insulin-like growth factor (IGF) pathway may play a role in the development and pathogenesis of colorectal cancer (5). Insulin is a promoter of colorectal neoplasia in animal models (6), and elevated circulating insulin and C-peptide levels (an indicator of insulin production) have been prospectively associated to colon cancer risk (7-9). The actions of insulin may in part be regulated through the IGF system (5). IGF-I and IGF-II regulate cell proliferation and differentiation and inhibit apoptosis. The availability of free IGF-I and IGF-II for interaction with their receptor (primarily IGF-I receptor) is modulated by the IGF binding proteins (IGFBP), especially IGFBP-3. IGFBPs can have opposing actions to IGF-I and IGF-II in part by preventing a ligand-receptor interaction (10) as well as through direct inhibitory and apoptotic effects on target cells (11). In prospective studies, relatively high plasma IGF-I (12-14) and IGF-II (15, 16) and low IGFBP-3 (16) levels were associated with greater risk of colorectal cancer.

An association between IGF proteins and QOL has been reported in several settings. In a study of 270 healthy patients, Unden et al. showed that IGF-I levels correlated with social well-being, mental well-being, depression, self-esteem, social support, self-rated health, and coping in a subgroup of middle-aged subjects (17). In a small randomized study of 33 obese postmenopausal women on a diet and exercise program, Thompson et al. found that women treated with IGF-I with or without growth hormone, but not growth hormone alone or placebo, had significant improvements in depression and anxiety scores (18). In contrast, Trojan et al. did not observe any associations between IGF-I levels and QOL measurements among 112 patients with post–polio syndrome (19). The mechanism of the potential influence of the IGF proteins on QOL is unclear, although the IGF/growth hormone axis has been correlated with improved organ functions, such as those of the brain (20) and heart (21, 22) as well as bone density (23), which may lead to enhanced physical and psychological well-being. To our knowledge, no such studies have examined the relationship between this important family of proteins and QOL in cancer patients.

A randomized treatment trial of patients with metastatic colorectal cancer was recently reported comparing irinotecan-based and oxaliplatin-based combination therapies (24). During the trial, we prospectively collected plasma samples and self-reported measurements of QOL. The effect of the different chemotherapies on QOL (25) and the relationship between baseline IGF protein levels and treatment efficacy (26) will be reported separately. In this report, we will focus on the association between pretreatment plasma levels of these biomarkers and baseline QOL as well as changes in quality over the course of therapy.

Patients included in this study were drawn from a national, intergroup randomized trial of chemotherapy for metastatic colorectal cancer [North Central Cancer Treatment Group (NCCTG) 9741; ref. 24]. Full details of the treatment trial have been published previously (24). Briefly, patients were required to have histologically proven unresectable colorectal adenocarcinoma; a baseline Eastern Cooperative Oncology Group performance status of ≤2; and adequate renal, liver, and bone marrow function. Exclusion criteria included prior therapy for advanced disease, baseline peripheral neuropathy or central nervous system disease, uncontrolled or severe comorbid illnesses, and baseline of more than three loose stools per day. Patients signed informed consent for participation in the trial and were given the option of inclusion in this companion study of plasma biomarkers. All patients were required to participate in the QOL portion of the study.

Blood samples were collected on study registration at respective institutions and sent to the Mayo Central Laboratory for Clinical Trials (Rochester, MN). Whole-blood samples were cooled and sent by overnight delivery to the laboratory. The stability of these biomarkers during the period of transport has been documented previously (27). Samples were centrifuged, divided, and frozen before use. IGF-I, IGF-II, IGFBP-3, and C-peptide levels were assayed in the laboratory of Dr. Michael Pollak using ELISAs with reagents provided by Diagnostic Systems Laboratory (Webster, TX). This methodology is more reproducible than and highly correlated (r = 0.98) with a RIA technique employed previously (28). Each patient had repeated measurements of each of biomarkers, with intraindividual correlations ranging from 0.98 to 0.99. Blinded quality control samples are included in each batch so that the mean intrapair coefficients can be assessed. In previous studies, the mean intrabatch coefficients of variation calculated from the quality-control samples were 7%, 5%, 9%, and 10%, respectively, for IGF-I, IGF-II, IGFBP-3, and C-peptide (9, 13, 29, 30).

The IGF ratio was defined as the molar ratio of the sum of IGF-I and IGF-II divided by IGFBP-3. The ratio serves as a surrogate for free or potentially bioavailable IGF-I.

Patients participating in the treatment trial were requested to complete several instruments measuring various aspects of patient QOL (Figs. 1 and 2). Protocol specifications were to obtain baseline QOL and then follow-up assessments every 12 weeks and at study completion for both instruments. The uniscale single-item instrument is a linear analogue measure of overall QOL that has been used in numerous clinical settings (31). It is valid and practical for patients with advanced colorectal cancer in community-based clinical trial settings (32). To obtain further details in other domains, the McCorkle and Young Symptom Distress Scale (SDS) was used (33). The SDS scale was developed for cancer patients and has been proven reliable and valid as well as prognostic for survival, although it has not been specifically tested in a cohort of only colorectal cancer patients (34-37). The SDS consists of 13 symptoms measuring degrees of distress related to appearance, appetite, breathing, bowel habits, cough, concentration, insomnia, fatigue, nausea frequency, nausea intensity, outlook, pain frequency, and pain intensity. For each symptom, five answers are possible (with a higher numerical response indicating a worse quality for that symptom). Due to reduced numbers of patient responses, the fourth and fifth lowest answers were collapsed into a single category. Each of the items was analyzed as separate individual constructs as well as a summated psychometrically validated scale, normalized to a scale of 0 to 100, with 0 being the worst QOL and 100 being the best QOL.

In total, 1,379 patients enrolled in NCCTG 9741 after the incorporation of an amendment to collect blood samples for companion biomarker studies. Of this cohort, 526 patients provided blood samples for these analyses; ∼80% of these patients completed at least portions of one of the two QOL. Baseline responses from the QOL instruments and changes in symptoms were studied in relationship to plasma levels of IGF-I, IGF-II, IGFBP-3, and C-peptide. Only patients who completed the baseline QOL instruments before starting chemotherapy were included in the baseline analyses of the uniscale and SDS scores; patients with missing values at baseline were excluded from that particular symptom score. Changes in symptoms were reported as the maximum decline or improvement from the baseline measurement over the course of participation in the trial; only patients completing the baseline instruments and at least one additional QOL assessment for an individual construct were included in each analyses. Trends for significance were measured by linear regression, with the dependent and independent variables treated as continuous variables. To satisfy the normality assumption, C-peptide was log transformed when modeling. Additionally, we adjusted analyses for age, gender, baseline performance status, and body mass index (BMI). For studies on IGF-I and IGF-II, we adjusted for IGFBP-3 and vice versa. Multivariate models were tested for collinearity and did not meet Belsey-Kuh-Welch criteria for collinearity. Further, for changes in symptoms, we adjusted for treatment arm. All statistical analyses used the SAS program package version 8.02 (SAS Institute, Cary, NC). Ps < 0.05 were used to denote statistical significance; all Ps reported are two sided.

We compared baseline characteristics of the overall cohort of patients who enrolled in the treatment trial with the subsets of patients participating in the biomarker studies and those completing QOL assessments (Table 1). We did not detect any appreciable differences between these subgroups. Further, patients experienced similar overall survival at a median follow-up of 20.4 months.

We examined the relationships among the IGF axis biomarkers, C-peptide, BMI, and baseline performance status (Table 2). We defined correlation coefficients of >0.6 as strong, between 0.3 and 0.6 as moderate, and <0.3 as weak or nonexistent (38). Using Spearman correlations, we found strong correlations among IGF-I, IGF-II, and IGFBP-3 but weak or no correlations for these proteins and either C-peptide, BMI, or performance status. C-peptide and BMI were marginally correlated with each other, whereas neither C-peptide nor BMI were associated with baseline performance status.

We examined the influence of initial BMI on baseline QOL. Whereas increasing BMI is predictive of worse QOL in noncancer populations (39, 40), lower BMI can also be predictive of increased cancer burden in patients (41, 42). In this cohort of patients with metastatic colorectal cancer, increasing baseline BMI was indeed predictive of improved perceptions of appearance, appetite, bowel function, and overall SDS score (Table 3). These findings remained unchanged after adjustment for age, gender, and baseline performance status. Baseline BMI was not predictive of subsequent changes in QOL while patients were treated on the clinical trial (Table 4).

We examined the association between baseline plasma levels of IGF-I, IGF-II, IGFBP-3, and C-peptide and baseline measures of QOL. We limited these analyses to patients who completed a QOL assessment before starting chemotherapy to understand the predictive value of these proteins toward the well-being of patients at presentation with metastatic disease. For each individual construct in the SDS, there were between 312 and 418 patients included. The median values of these proteins based on their response to the instruments are presented in Table 5. For the total SDS and uniscale scores, patients were divided into quartiles based on their responses.

In unadjusted analyses, higher levels of IGF-I were significantly related with improved QOL related to appearance, appetite, breathing, bowel function, fatigue, nausea frequency and intensity, and uniscale and total SDS. After adjustments for age, gender, IGFBP-3, and baseline performance status, the associations with appearance, appetite, and nausea intensity remained statistically significant. Additionally, after multivariate regression, distress from cough symptoms was also associated with IGF-I. These relationships remained significant when BMI was added into the model.

Similarly, IGF-II levels were associated with perceptions of appearance, appetite, breathing, bowel function, concentration, fatigue, nausea frequency and intensity, pain frequency, and uniscale score and total SDS. Appearance, appetite, fatigue, nausea frequency and intensity, pain frequency, and total SDS remained significant after adjustment for the above-listed potential confounders. The IGF ratio, incorporating IGF-I and IGF-II and adjusted for IGFBP-3 binding, showed similar results to IGF-II, with positive associations to appearance, appetite, breathing, fatigue, nausea frequency and intensity, and SDS composite score.

In contrast, IGFBP-3 and C-peptide were not associated with any of individual constructs or cumulative scores particularly after adjustments for potential confounders.

We considered the possibility that the association between plasma IGF-I and IGF-II and baseline QOL measures may simply reflect either individual patient disease burden or nutritional status. We therefore repeated our analyses after excluding patients with a baseline BMI of <21 kg/m² or a baseline performance status of 2. Among the remaining 347 patients, the associations between plasma IGF-I and IGF-II and baseline QOL measures remained essentially unchanged, albeit attenuated due to decrease in sample size. In this “healthier” cohort, the adjusted test for trend remained significant between IGF-I and appearance (P = 0.02), appetite (P = 0.01), and cough (P = 0.05) and between IGF-II and appearance (P = 0.006), appetite (P = 0.002), fatigue (P = 0.04), nausea frequency (P = 0.001), nausea intensity (P = 0.04), and SDS composite (P = 0.005).

We examined whether these baseline biomarkers would predict subsequent changes in QOL during treatment for metastatic disease. As a primary analysis, we examined maximum decline in each of the QOL constructs from baseline, limiting the analyses only to patients with at least two measurements. None of the biomarkers nor BMI predicted declines in QOL as measured by the SDS composite and uniscale (Table 4). Similarly, there were no associations between baseline IGF-I, IGF-II, IGFBP-3, IGF ratio, C-peptide, or BMI and any of the individual SDS constructs (P > 0.05). Similarly, no associations with improvements in QOL were detected (data not shown). Finally, there were no relationships between the IGF proteins or C-peptide and either declines or improvements of at least 10 points in the uniscale or SDS composite score within the same patient, which have been considered to be clinically meaningful (ref. 43; P > 0.05, all χ²).

Using a cohort of patients with previously untreated metastatic colorectal cancer, we showed that lower levels of plasma IGF-I and IGF-II are associated with greater baseline symptom distress before initiation of chemotherapy. In contrast, baseline levels of IGFBP-3 and C-peptide were not associated with measures of QOL. None of the baseline biomarkers were associated with subsequent declines or improvements of symptoms while being treated with first-line chemotherapy. Finally, higher baseline BMI was associated with improved baseline QOL and diminished symptom distress.

Several potential mechanisms may explain these findings. IGF-I and IGF-II may have direct beneficial biological effects on patients with cancer that lead to improved indicators of QOL. Indirect support of this hypothesis is seen in patients with growth hormone deficiency treated with growth hormone replacement, where IGF-I levels increase and QOL improves (44, 45). Administration of recombinant IGF-I improved QOL in certain chronic disease states (46), although not others (47, 48). Conversely, plasma levels of IGF-I and IGF-II may be a reflection of disease burden and nutritional status that are a consequence of disseminated cancer. IGF-I, IGF-II, and C-peptide levels are influenced by malnutrition (49-52). Notably, our findings for IGF-I and IGF-II did not change after controlling for baseline BMI and did not change after excluding leaner patients and those with an impaired performance status. Further, although C-peptide levels have been associated with nutritional status (53, 54) C-peptide was not associated with any QOL measure in the current analysis.

Preliminary results from this cohort of patients with metastatic colorectal cancer show that higher IGFBP-3 levels are associated with increased likelihood of tumor response to chemotherapy and longer time to progression and overall survival (26). In contrast, IGF-I was not correlated with response rate, but increased levels were correlated with prolonged time to progression and overall survival. We believe our data support these findings, in that higher IGF-I may be reflective of a “better host” (55, 56) who experiences superior indicators of QOL, whereas IGFBP-3 may have an independent effect beyond the state of the patient. This explanation is purely speculative and will require further study in colorectal cancer.

Our observation that increased BMI was associated with an improved QOL likely reflects the fact that cancer patients with lower BMIs may have suffered weight loss before diagnosis, which is predictive of worse outcomes in colorectal cancer (57). Additionally, other studies have shown that obese patients have fewer side effects from chemotherapy (58, 59), which may be a consequence of their improved QOL at baseline.

Several limitations should be noted in this study. Although baseline BMI was recorded in this study, we did not collect information on weight loss that may have occurred immediately before study enrollment. Thus, residual confounding by cachexia associated with the diagnosis of metastatic colorectal cancer cannot be excluded. In addition, both protein levels and QOL may be affected by other physical and psychological factors affecting on individual patients before enrollment in a clinical trial, such as morbidity from recent surgery and physical symptoms related to the location of metastases. Although inclusion of performance status in our regression model may partially adjust for such differences, such unaccounted factors may not be adequately controlled for in our analyses. Protein levels were only measured at the start of chemotherapy, and the effect of change over time of these growth factors on QOL could not be assessed. However, one study of women receiving chemotherapy for advanced breast cancer showed that IGF-I did not change on therapy, whereas IGFBP-3 only modestly decreased (60).

As with other QOL studies, missing data can bias results. In our primary analyses of baseline IGF proteins and C-peptide, we only included patients who completed the QOL instruments before starting chemotherapy. However, if sicker patients with reduced QOL did not complete the survey, our results would be biased toward the null and we find multiple positive associations. Additionally, no differences in survival were observed for patients who did or did not complete the QOL assessments and provided blood sample. Finally, multiple comparisons can lead to positive relationships by chance. In the baseline analyses, we examined six markers against 15 QOL constructs (90 comparisons). By chance alone, one might anticipate four to five positive associations; however, we showed 22 significant associations. Moreover, no associations were found with either C-peptide or IGFBP-3, which could be expected by random chance.

We did not detect a difference in declines or improvements of QOL based on any of the protein levels. These analyses may be limited by differential completion of follow-up QOL assessments, which may have biased the results toward the null. However, other such longitudinal studies in cancer patients have also had decreasing compliance with completion of follow-up instruments and were able to show associations between outcome and changes in QOL (61-64). Moreover, because we measured IGF proteins only at baseline, we were not able to add potential changes in those levels in our longitudinal models.

There is a growing interest in the IGF pathway as a potential target for cancer therapy (65). Although these data cannot address the merits of such a strategy, we would advocate close observation of QOL when testing such agents.

Genetic variants in folate metabolism were recently shown to be associated with QOL indicators in patients with metastatic colorectal cancer (4). We believe that these findings, in conjunction with the current study, begin to provide a link between molecular biology and QOL research. There is a strong interest to identify molecular biomarkers and genetic polymorphisms to tailor drugs for individual patients. In addition to predicting prognosis and response to therapy, such biological markers may also allow clinicians to incorporate QOL considerations when deciding on a course of treatment.

We thank Erin M. Green for her assistance with data management and Paul J. Limburg, M.D., for his helpful review of the article.