A prospective study to investigate the role of serial serum mesothelin in monitoring mesothelioma

Malignant pleural mesothelioma (MPM) is an aggressive malignancy with a median survival of 9–14 months [1]. There has been little improvement in overall survival from the disease over the last decade with limited development of effective treatments. Chemotherapy is the only treatment modality that has shown to improve survival following the results of a large randomised control trial (RCT) of pemetrexed and cisplatin which conferred an additional 2 months to overall survival compared to cisplatin alone [2]. Due to a response rate of only 25% and a significant side effect profile many patients decline or are unsuitable for first line chemotherapy [3]. The role of non-chemotherapeutic options such as biological treatments or immunotherapy is the topic of many ongoing clinical trials [4‐8]. Additionally, the efficacy of second line chemotherapy in patients who have progressed following standard chemotherapy is uncertain with promising reports from the use of vinorelbine, gemcitabine or re-challenging with pemetrexed [9‐12].

Given the low response rate to conventional treatment, and uncertainty around the efficacy of newer therapies, the ability to accurately monitor disease is essential for oncologists and trialists. The current standard for monitoring is, as for many tumours, serial computed tomography (CT) scans. However, unlike many tumours, MPM does not grow spherically, instead growing as a ‘rind’ encasing the thoracic cavity. An adaptation of the Response Evaluation Criteria in Solid Tumours (RECIST) to the modified RECIST criteria has only partially allowed for this unique morphology. Other issues including low sensitivity to predict progression, subjectivity in radiologists’ choice of measurement sites and complications from the presence of pleural plaques or fluid, mean research into a more robust and reproducible radiological marker is ongoing [13]. Studies have shown that serial FDG-PET (Fludeoxyglucose positron emission tomography) scanning can be used to predict response [14] but is significantly limited by false positive uptake in inflamed tissue from surgical intervention or pleurodesis, as well as restricted availability or limited reimbursement from insurers [15].

Serum or pleural fluid biomarkers that could reflect disease activity are an attractive option for MPM given the difficulties with radiological assessment. The most investigated is mesothelin, a membrane-bound glycoprotein overexpressed by malignant mesothelial cells [16]. Soluble mesothelin is found in the blood and pleural fluid of patients with MPM and levels correlate with tumour stage and bulk. It lacks the sensitivity to be used as a diagnostic marker given reduced expression in non-epithelioid MPM. There remains uncertainty as to whether the baseline mesothelin level is a useful predictor of prognosis. However, a systematic review we performed found 8 studies that had assessed serum mesothelin’s ability to monitor disease during chemotherapy and found it correlated with radiological markers and survival [17]. No studies had assessed the utility of mesothelin in longer term monitoring patients not currently receiving chemotherapy. In this prospective cohort study, we aim to assess the ability of serum mesothelin to monitor disease at later timepoints in patients who have completed chemotherapy or those receiving best supportive care (BSC).

This study assessed the ability of serum mesothelin to detect progression of MPM in patients following chemotherapy or receiving best supportive care. To our knowledge this is the first study to focus on longer term monitoring as opposed to mesothelin’s role as a marker of treatment response during or immediately after chemotherapy. The ability to monitor MPM in these patients is becoming increasingly important. However, no specific evidence exists for how patients with MPM should be monitored following first line chemotherapy. Recent guidelines have suggested that 3–4 month follow up is good practice without specifying what form this follow up should take https://​www.​brit-thoracic.​org.​uk/​standards-of-care/​guidelines/​bts-statement-on-malignant-mesothelioma/​. There is considerable variation in practice nationally, for two main reasons. Firstly, until recently many oncologists felt there was no data to support the role of second line chemotherapy in this setting. In addition, there was limited access to clinical trials after first line treatment. There was, therefore, no perceived benefit from monitoring patients post first line chemotherapy or in those having BSC. However, given promising results from second line chemotherapy trials, as well as non-chemotherapeutic options this perception is changing. Secondly, no radiological marker has shown the ability to accurately monitor disease. A serum biomarker that could identify disease progression would be very useful to oncologists and physicians, with the additional benefit of reducing patient burden and healthcare costs. Mesothelin is the most studied serum biomarker for mesothelioma with the majority of literature assessing its diagnostic potential. Although a raised level is fairly specific for MPM, many patients with non-epithelioid disease will have unrecordable levels even at an advanced stage. This has limited its utility as a diagnostic test, although recent studies have attempted to combine it with more novel markers [18, 19]. The role of biomarkers in other malignancies has often begun as a putative diagnostic marker before becoming a marker of treatment response or recurrence (CA125, PSA).

Previous literature has shown that mesothelin can be used as a marker of treatment response when measured serially. The first example from Grigoriu and colleagues [20] took patients with a positive (> 1 nM/L) mesothelin at baseline receiving chemotherapy and immunotherapy (n = 40). They found that in patients with a 10% or greater rise in mesothelin there was a 75% chance of radiological progression using mRECIST as well significantly worse overall survival. The largest study of its type [21] correlated both CT scans (mRECIST) and PET (TGV and tumour volume) with change in mesothelin levels after treatment (using a 25% cut-off). In the chemotherapy group (n = 55) the change in mesothelin significantly correlated with radiological response on CT (p = 0.023). Although only a minority of the cohort (n = 28) could be reassessed using PET imaging, due to prior pleurodesis or surgery, the change in both metabolic activity (TGV) and tumour bulk strongly correlated with change in mesothelin levels (p < 0.001). Survival analysis demonstrated that the trend of mesothelin correlated with survival in a multivariate model that included age, sex, histology and treatment. Interestingly, this was not true when tumour volume on PET was added to the model, indicating that mesothelin was probably acting as a proxy for tumour bulk.

Across the literature we found several different thresholds for defining a clinically significant change in mesothelin level, including 10% [20], 15% [22] and 25% [21]. Wheatley-Price and colleagues [23] performed a post-hoc analysis of mesothelin values from a cohort of 42 patients using absolute (5 mmol) and relative (10%) changes and their correlation with survival, finding that relative changes were more accurate. This current study assessed a variety of cut-offs, finding that a 10% cut-off reduced the specificity of a rising mesothelin level but had a sensitivity of 96%. Given our aim was to assess the ability of mesothelin to detect disease progression we felt this reduction in specificity was justified with a false positive rate of 28% but a false negative rate of only 4%.

We demonstrated that disease progression could be detected using mesothelin with an accuracy of 86% and NPV of 94%. Unlike some previous literature [20, 23] this analysis did not exclude patients from the analysis with low baseline mesothelin or non-epithelioid disease. There has been uncertainty about the utility of serially measuring mesothelin in these patients. In this study, 44% (8/19) of patients with a mesothelin of less than 2 mmol/L at baseline had an increased level (range 2.1–20.4) at later timepoints. Additionally, when the analysis was limited to patients with a baseline mesothelin of < 2 mmol there was only a limited reduction in sensitivity to predict progression. Given the small subgroup, no firm recommendations can be made regarding the utility of retesting mesothelin when the baseline is negative (< 2 nmol/l). This needs further assessment in other prospective studies.

This study had several limitations that could impact on the conclusions drawn. The cohort of patients was small, which has an impact on the conclusions of the multivariate analyses, but comparable to other studies of its type. This was an observational study alongside normal clinical practice and there were instances where a mesothelin was collected but the patient did not receive a scan within the pre-allocated 1 month period. Additionally, given the aggressive nature of MPM very few comparisons of timepoints over 12 months were available, but this is likely to be a pragmatic assessment of following up these patients in clinical practice. As previously mentioned other studies have monitored the change in mesothelin levels in response to systemic therapy. As a result, these studies had a proportion of patients where mesothelin levels fell considerably, which was correlated with radiological ‘partial response’. Given this study focused on patients not receiving treatment there were few instances of a falling mesothelin. Worsening renal function has been shown to falsely elevate mesothelin levels [24]. Although there was no difference in mesothelin sensitivity between patients with normal and abnormal renal function at baseline, there was no ongoing assessment during follow up. Finally, the difficulty with studies of this type is that the current ‘gold standard’ of monitoring (mRECIST CT) which mesothelin trends are compared to has been shown to be insensitive at detecting early progression. This analysis found that mesothelin correlated better with the clinically reported CT scan compared to mRECIST reporting, a finding of other similar studies [23]. Additionally, over half (25/43) of the CT scans used in the timepoint analysis were non-measurable by mRECIST criteria with a short axis diameter of less than 1 cm. This is another shortcoming of the mRECIST criteria that limits its sensitivity and applicability in clinical practice. It is for this reason that we assessed the impact of a changing mesothelin levels on survival as a means of validation, finding that a rising mesothelin at 6 months was a poor prognostic indicator. As with other cancer biomarkers [25] the reason for several ‘false-positive’ results (where a rising mesothelin occurred in the context of radiologically stable disease) could be because changes in mesothelin preceded radiological change.

We have demonstrated that a rising serum mesothelin is a sensitive marker of progression in the follow up of patients with MPM. Given the emergence of effective non-chemotherapeutic treatments and second-line agents, accurate disease monitoring is becoming increasingly important. This study has demonstrated that regular mesothelins could be used as a cost effective adjunct or alternative to serial CT scanning.