Calretinin as a blood-based biomarker for mesothelioma

Abstract

Background: Malignant mesothelioma (MM) is a deadly cancer mainly caused by previous exposure to asbestos. With a latency period up to 50 years the incidence of MM is still increasing, even in countries that banned asbestos. Secondary prevention has been established to provide persons at risk regular health examinations. An earlier detection with tumor markers might improve therapeutic options. Previously, we have developed a new blood-based assay for the protein marker calretinin. Aim of this study was the verification of the assay in an independent study population and comparison with the established marker mesothelin.

Methods: For a case-control study in men, a total of 163 cases of pleural MM and 163 controls were available from Australia, another 36 cases and 72 controls were recruited in Germany. All controls had asbestosis and/or plaques. Calretinin and mesothelin were determined by ELISA (enzyme-linked immunosorbent assay) in serum or plasma collected prior to therapy. We estimated the performance of both markers and tested factors potentially influencing marker concentrations like age, sample storage time, and MM subtype.

Results: Calretinin was able to detect all major subtypes except for sarcomatoid MM. Calretinin showed a similar performance in Australian and German men. At a pre-defined specificity of 95% the sensitivity of calretinin reached 71% and that of mesothelin 69%, when excluding sarcomatoid MM. At 97% specificity, the combination with calretinin increased the sensitivity of mesothelin from 66% to 75%. Sample storage time did not influence the results. In controls the concentrations of calretinin increased 1.87-fold (95% CI 1.10-3.20) per 10 years of age and slightly more for mesothelin (2.28, 95% CI 1.30-4.00).

Conclusions: Calretinin could be verified as a blood-based marker for MM. The assay is robust and shows a performance that is comparable to that of mesothelin. Retrospective analyses would not be limited by storage time. The high specificity supports a combination of calretinin with other markers. Calretinin is specific for epithelioid and biphasic MM but not the rarer sarcomatoid form. Molecular markers like calretinin and mesothelin are promising tools to improve and supplement the diagnosis of MM and warrant further validation in a prospective study.

Keywords: Asbestos; Biomarker panel; Biphasic; Calretinin; Early diagnosis; Epithelioid; Mesothelin; Mesothelioma; Plasma; Sarcomatoid; Serum.

Fig. 1

Marker concentrations in MM subtypes. a Calretinin [ng/mL] in controls and MM cases by subtype. b Mesothelin [nmol/L] in controls and MM cases by subtype. All cases and controls were from Australia (group 1). Individual p-values relate to the comparison between each subtype and the controls. P-values for calretinin were obtained from two-sided Peto-Prentice test and for mesothelin from two-sided Wilcoxon rank-sum test

Fig. 2

Comparison of marker concentrations in samples from Australia and Germany. a Calretinin [ng/mL] in MM cases and controls from Australia (group 1 and 2) and Germany (group 3). The corresponding p-values (group 2 vs. group 3) are: p = 0.8210 for MM cases and p = 0.0773 for controls. b Mesothelin [nmol/L] in MM cases and controls from Australia (group 1 and 2) and Germany (group 3). The corresponding p-values (group 2 vs. group 3) are: p = 0.0012 for MM cases and p = 0.1422 for controls. P-values for calretinin were obtained from two-sided Peto-Prentice test and for mesothelin from two-sided Wilcoxon rank-sum test. For better comparison, for group 1 sarcomatoid MM were excluded

Fig. 3

ROC analyses of calretinin and mesothelin in samples from Australia and Germany. a Nonparametric (AUC = 0.90, 95% CI = 0.85–0.95) and bi-lognormal (AUC = 0.95, 95% CI = 0.92–0.98) ROC curves for calretinin in Australian samples (group 2). b Nonparametric (AUC = 0.83, 95% CI = 0.74–0.92) and bi-lognormal (AUC = 0.87, 95% CI = 0.79–0.95) ROC curves for calretinin in German samples (group 3). c Nonparametric (AUC = 0.91, 95% CI = 0.87–0.96) and bi-Weibull (AUC = 0.93, 95% CI = 0.90–0.96) ROC curve for mesothelin in Australian samples (group 2). d Nonparametric (AUC = 0.84, 95% CI = 0.76–0.93) and bi-Weibull (AUC = 0.85, 95% CI = 0.81–0.89) ROC curve for mesothelin in German samples (group 3)

Fig. 4

ROC analyses of calretinin and mesothelin with pooled data from Australia and Germany. a Nonparametric (AUC = 0.86, 95% CI = 0.82–0.91) and bi-lognormal (AUC = 0.90, 95% CI = 0.86–0.94) ROC curves for calretinin. b Nonparametric (AUC = 0.89, 95% CI = 0.85–0.93) and bi-Weibull (AUC = 0.91, 95% CI = 0.89–0.94) ROC curves for mesothelin. All ROC curves are based on pooled data from group 2 and 3

Fig. 5

Scatterplot of calretinin versus mesothelin. The plot shows marker concentrations of MM cases and controls from Australia (group 2) and Germany (group 3)

Fig. 6

Scatterplot of calretinin versus storage time. Marker concentrations [ng/mL] in MM cases and controls from Australia and Germany (group 1, 2, and 3) were plotted against storage time [months]

Fig. 7

Scatterplot of marker concentrations versus age. a Concentrations of calretinin [ng/mL] were plotted against age [years] of MM cases and controls. b Concentrations of mesothelin [nmol/L] were plotted against age [years] of MM cases and controls. The plots are based on pooled data from group 2 and 3