Background
Diabetes has been associated with an increased incidence and mortality in many types of cancer [
1‐
3]. Diabetes may influence cancer progression and outcome by several mechanisms, including hyperinsulinemia, hyperglycemia, or chronic inflammation. In addition, type 2 diabetes mellitus (T2DM) is a risk factor for other non-neoplastic causes of death in cancer patients [
4].
The role of diabetes in the prognosis of patients with lung cancer has not been well established [
5]. Some clinical and epidemiologic studies showed that pre-existing diabetes has a negative impact on lung cancer mortality [
5‐
7]. A worse prognosis in diabetic patients with lung cancer has been mainly seen in women [
6], patients with non-small cell lung cancer (NSCLC) and in those patients treated with surgical resection [
5]. However, these observational studies have methodological limitations such as the absence of screening for T2DM [
2], the lack of cause-specific death reporting and the heterogeneity of tumor stage at cancer diagnosis [
7].
The use of baseline glycemia as a continuous variable could avoid the problem of T2DM underdiagnosis. In fact, fasting plasma glucose (FPG) level has been independently associated with an increased risk of mortality in a cohort of patients with newly diagnosed NSCLC [
7]. In addition, The Emerging Risk Factors Collaboration Study analyzed the potential independent associations between diabetes or hyperglycemia with the risk of death from cancer. This study showed an association between high FPG levels (exceeding 100 mg per deciliter or 5.6 mmol per liter)and cancer specific mortality [
8].
The influence of antidiabetic therapy on cancer incidence and mortality has been the subject of many observational epidemiologic studies in large populations [
9,
10]. These studies suggest that metformin is associated with a decrease in cancer mortality, while data regarding other antidiabetic medications are conflicting [
10,
11]. The Danish Cancer Registry provided evidence that among cancer patients with pre-existing diabetes, mortality rates were higher on patients receiving insulin [
12].
The purpose of our study is to determine the influence of T2DM and FPG levels on survival in a cohort of patients with unresectable locally advanced NSCLC treated with concurrent chemoradiotherapy. The influence on overall survival (OS) of antidiabetic treatments and glycemic control among T2DM patients were also evaluated as secondary endpoints.
Discussion
In this homogeneous cohort of consecutive patients with unresectable locally advanced NSCLC, both pre-treatment FPG level and diagnosis of T2DM were predictors of overall survival. However, only FPG level retained its significance in the multivariate analysis. Interestingly, PFS, a cancer-related outcome, was significantly shorter for patients with high FPG, suggesting a direct relationship between both variables.
Several mechanisms may explain the negative effect of hyperglycemia and diabetes on lung cancer related outcome. Hyperinsulinemia and metabolic rewiring of cancer cells are potential factors contributing to the development and progression of cancer. According to Warburg’s hypothesis, the hyperglycemic environment may accelerate the proliferation of cancer cells, as they can obtain essential metabolites and energy mainly from fermentation of glucose, even in aerobic conditions [
16,
17]. Data from patients and mouse models indicates that lung tumors are dependent on glucose metabolism, and increased expression of glycolytic enzymes correlate with poor prognosis. In our study, squamous cell carcinoma histology was more probable in diabetic patients. A recent case-control study showed an increased risk of lung cancer associated with glycemic load and dietary glycemic index, as a marker of carbohydrate intake and postprandial glucose response. The risk was greater for the development of squamous cell carcinoma [
18]. In another study, squamous cell lung carcinoma was associated with enhanced glucose uptake and exhibited higher glycolytic dependency than adenocarcinoma [
19].
Han et al. found that high glucose levels can promote cancer proliferation via the induction of epidermal growth factor (EGF) expression and transactivation of EGF receptor [
20]. Hyperglycemia has also shown to decrease the antiproliferative effect of chemotherapy in preclinical models, but the results are inconsistent and have not been proven in clinical trials [
21]. Some retrospective series have also shown a negative impact on survival of elevated blood glucose levels during radiation therapy in glioblastomas. This detrimental effect may be explained by the metabolic impact on tumor microenvironment, but also by the induction of hypoxia that promotes resistance to radiotherapy [
22,
23].
The association of high FPG with outcome in lung cancer patients in a clinical setting has already been reported. In an unselected cohort of 342 patients with newly diagnosed NSCLC, Luo et al. observed that patients with FPG levels ≥7.0 mmol/L (≥ 126 mg/dl) had shorter survival outcome independently of other prognostic factors [
7]. In another retrospective study, 159 patients with locally advanced NSCLC were treated with radical chemoradiotherapy and those patients who had a previous diagnosis of diabetes or with FPG ≥ 7.0 mmol/L had a significantly shorter survival independently of other prognostic factors [
24]. In this study, FPG level was not studied in a separate model in the multivariate analysis, and the value of this parameter at baseline as a continuous variable was not clarified. By contrast, in our study, previous diagnosis of T2DM lost its statistical significance in the multivariate analysis, whereas FPG maintained a strong effect.
To evaluate the influence of metabolic control of diabetes on cancer outcome, we explored the association of HbA1c (as a dichotomous variable) with the prognosis of lung cancer. We observed that diabetic patients with the poorest glycemic control (HbA1c > 8.5%) had the shortest survival, and that diabetic patients with good metabolic control (HbA1c < 7%) had a similar OS than non-diabetic patients. Moreover in a multivariate model poor metabolic control was independently related to prognosis. These findings underline the potential benefit of achieving good metabolic control to improve cancer outcome in diabetic patients [
25].
The influence of antidiabetic treatment on cancer prognosis is still controversial. Metformin is an antidiabetic drug that suppresses hepatic neoglucogenesis and improves insulin sensitivity. Preclinical data have shown that metformin has direct and indirect potent antitumor effects in several cancers, including lung. However the concentration of metformin used in these studies is about one thousand times higher than that achieved in the blood in humans [
26]. Some epidemiological studies in lung cancer have shown better outcomes in metformin-treated patients compared to diabetic patients receiving other antidiabetic drugs [
10,
11,
27]. Ahmed I et al., in a study with 166 patients, did not find any differences in survival between diabetic and non-diabetic patients except for being on metformin treatment or not [
28]. More than 200 clinical trials [
10] are ongoing aiming to determine whether the preclinical anticancer effect of metformin translates into clinical benefit. The Danish Cancer Registry showed a higher mortality among patients treated with insulin [
12]. In line with those reports, we observed that patients treated with insulin (± oral hypoglycemic drugs) had a worse prognosis than patients on metformin treatment (without insulin) or those that did not receive any antidiabetic treatment.
A common concern when studying the relationship between diabetes and cancer prognosis is the higher prevalence of comorbidities in diabetic patients and the existence of competing causes of death. We did not identify significant differences in the cause of death between diabetic and non-diabetic patients. Although patients with T2DM had higher prevalence of overweight, body mass index (BMI) was not a significant prognostic factor in our study. We also collected basic malnutrition measures as weight loss before treatment and baseline albumin level as known prognostic factors. However, none of these factors was different between the T2DM and the non-diabetic patients or achieved significance in the survival analysis. Another prominent issue is that the presence of T2DM could lead to a less intensive anticancer treatment. Although diabetic patients were more likely treated with carboplatin than cisplatin and, carboplatin treatment was associated with unfavorable overall survival in the multivariate analysis, baseline FPG, metabolic control and insulin treatment remained independent prognostic factors after adjusting by platinum treatment.
The major limitation of our study is the relatively low sample size, which could lead to residual confounding as we do not have enough power to identify some well-known risk factors and compromises the analysis of subsets of patients. Another weakness is the retrospective nature of the study, which means that there was not a dedicated pre-planned design of patient management and data collection. The strength of our study is the uniform clinical practice on cancer and diabetic management at a single institution. The most relevant finding is that baseline FPG level is a strong predictor of survival in a set of consecutive patients treated with definitive chemoradiotherapy. Our results on the prognostic value of metabolic control and antidiabetic treatment have potential clinical implications and are consistent with what has been reported in the literature, but should be regarded with caution due to the limitations of the study.