Personalized Chemosensitivity Assays for Mesothelioma: Are They Worth the Effort?

In this issue of Clinical Cancer Research, Schunselaar and colleagues “chemically” profiled primary mesothelioma cultures with common chemotherapeutic drugs to determine, in real time, sensitivity and resistance patterns in addition to genomic characterization (1). Since the first human cancer cell line was cultured in 1951, the translatability of short- and long-term cell cultures has been questioned as representative paragons of the tumors they were set up to model. The inception of the NCI-60 panel did enable researchers to develop further biological hypotheses for drug sensitivity and resistance; however, the ability to predict a priori response has only more recently been tested in short-term cultures and in vivo models with variable success (2). One of the main issues with this approach is the labor intensity required to obtain, disaggregate, and then culture tumors from biopsies and cellular fluid. Second, the translatable dose from in vitro applications into in vivo models and indeed humans is not well defined, especially in mesothelioma. Lastly, tumor heterogeneity, immune cells, and physical drug access barriers are lacking with in vitro systems but are of major importance in clinical practice (Fig. 1).

Despite these barriers, the authors' determination to implement their testing strategy is to be applauded. Not only were they able to obtain and culture pleural fluid from 102 patients with a rare tumor, but they developed short-term cultures from 155 samples and performed drug screens in over half. However, it is important to realize that despite their extensive efforts, chemical profiling was available in only 57 out of 102 patients, mainly due to low tumor cell count in 45 samples. The stringency of their cell lines was tested using array cGH to ensure that the cells within the culture were mesotheliomas genomically—a very important quality control, as overgrowth of cell cultures with fibroblasts and mesothelial cells is common.

More information about the clinical correlation of chemical profiling from the 57 patients would better define the utility of these assays. Instead, correlative data between response prediction and actual clinical response were available for only 10 cases. Seven of 11 drug screens from 10 patients were correctly predicted by the assays; however, whether the prediction of disease stabilization is valid is debatable given that the difference between response and stabilization in vitro occurred over a large dose range and between 25% and 50% of cell kill. Stabilization was the best response in five of 11 lines, and the assays only predicted two samples to be sensitive, and this was correct in only one. It would be a lot more interesting to know how accurate the preclinical predictions were for the remaining cell lines, as it is not possible to draw any conclusions from the small sample size provided. It is interesting that the authors chose unusual drug combinations to test, such as oxaliplatin and vinorelbine, a regimen rarely used clinically. Although this is understandable when searching for further treatment strategies, greater proof of principal could be established if they had considered response to platinum in those 41 patients whose tumors were treatment naive and compared this with the samples that would have received platinum-based treatment upfront. Unfortunately, there are limitations to the use of pemetrexed, the commonest cytotoxic used in mesothelioma, in both in vitro and in vivo models, which the authors discuss in detail. It is important to define patients likely to respond to specific cytotoxics, especially where most are unlikely to benefit. For example, single-agent, second-line vinorelbine yielded an objective response rate (ORR) of only 16%, with overall survival (OS) of 9.6 months. Similarly, combinations of gemcitabine and vinorelbine have shown an ORR of 10%, with OS of 10.9 months (3). The ability to detect the one in 10 likely to benefit promises to deliver the best therapy to these patients while sparing those unlikely to benefit.

Drug sensitivity is a complex interplay between the ability of a cytotoxic agent not only to perturb cellular functions but also to obtain cellular access via the vascular compartment through the tumor stroma. The immunologic milieu clearly plays a role, considering the improved efficacy of combination chemoimmunotherapy over either modality alone in several tumor types. The location of mesothelioma in pleural surfaces and an often-inflamed stroma hinders such cellular access and may result in discordance between in vitro and in vivo response. Certainly, in other tumor types, such as lung cancer and ovarian cancer, no consistent correlation has been observed between in vitro and clinical response, and the utility of these strategies has therefore not gained traction (4).

More recently, immune checkpoint inhibitors and antiangiogenic agents have shown efficacy in mesothelioma. In vitro chemical profiling may be able to determine whether an antiangiogenic agent kills tumor cells, but its real effect on the tumor vasculature cannot be modeled using these assays. Similarly, the immune microenvironment is obviously absent, meaning such approaches cannot be investigated.

The most important utility of the reported cell culture system is to identify putative pathways of resistance using molecular profiling. Toward this end, the authors defined the response profiles of 81 cell lines and grouped them into responders and nonresponders. Perhaps surprisingly, the profiles appeared not to cluster according to the treatment naivety of the cells. More importantly, upregulation of the FGF pathway in a nonresponder patient population was shown to be present and targetable by FGF inhibitors. Potentially, a combination of biomarker analysis using an in vitro cell culture system could be an important initial step in the individualization of anticancer therapy on progression with standard therapies.

Clinical research for mesothelioma has often been hampered by the rarity of the disease, and few positive trials have resulted in a decade without any new agents entering standard therapeutic paradigms. More recently, promising signals from trials using angiogenesis inhibitors and immunotherapy offer a potential light at the end of a long tunnel. Although there is much excitement about novel therapies and targets, there remains room to improve on patient selection for chemotherapy—the mainstay of treatment for most patients with mesothelioma. These data provide evidence that personalized cell culture models could aid in such decision-making; however, without a lot more validation of the predictability of the assays, one wonders whether the intensive nature of such a protocol is really worth the effort. Adoption of such models would require the demonstration of an improvement in not only response to therapy but also OS, an endpoint that few other studies in other tumor types have been able to achieve (5).

No potential conflicts of interest were disclosed.

Conception and design: T. John

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. John

Writing, review, and/or revision of the manuscript: T. John, P.L. Chia

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. John

T. John received an NHMRC Early Career Fellowship. P.L. Chia received an International Association for the Study of Lung Cancer (IASLC) Fellowship Award and The University of Melbourne, Australian Postgraduate Award and was supported by the Lyall Watt's Mesothelioma grant awarded by Cancer Council Victoria and Victorian Cancer Agency.