VeriStrat: a prognostic and/or predictive biomarker for advanced lung cancer patients?

Abstract

The role of EGF receptor (EGFR) inhibitors in the treatment of lung cancer without activating EGFR mutations has been a controversial issue, particularly their relative efficacy over the available chemotherapy in the second- and third-line setting. VeriStrat is a serum/plasma proteomic test developed using matrix-assisted laser desorption/ionization methodology, aiming at predicting benefit from EGFR treatment. The VeriStrat algorithm has been interrogated retrospectively and prospectively in samples from randomised trials, such as the PROSE study, confirming the prognostic information associated with the signature. In addition, the test appeared to be predictive of erlotinib impact on survival, as only VeriStrat Good patients benefited from such a treatment. Additional studies should confirm and further define its role in predicting EGFR tyrosine kinase inhibitor benefit, and to establish its better use in terms of clinical efficiency identifying which patients are candidates for the test, at which time on the history of the disease, and lastly at what extra cost.

Keywords::

• advanced NSCLC
• EGFR
• proteomic signature
• serum
• tyrosine kinase inhibitors
• VeriStrat

Treatment alternatives for NSCLC beyond first line

Lung cancer is one of the most common malignant tumors and the leading cause of cancer-related death worldwide, being nonsmall cell lung cancer (NSCLC) the most frequent histological subtype [1]. Despite attempts to improve early detection methods, the vast majority of NSCLC patients are diagnosed with advanced unresectable stage disease and are typically candidates for systemic therapy. First-line therapy in advanced NSCLC include several platinum-based chemotherapy regimens [2]. The taxane docetaxel and the antimetabolite pemetrexed are widely accepted treatment alternatives in the second-line setting [2,3]. In addition, the EGFR tyrosine kinase inhibitors (TKIs) erlotinib and gefitinib have proven activity after one or two prior chemotherapy treatments. Of relevance, the NCIC BR21 trial showed the superiority of erlotinib as compared with placebo in unselected patients who have received two chemotherapy treatments or were not candidates for second-line chemotherapy (overall survival [OS] hazard ratio [HR]: 0.70; p < 0.001) [4]. Additional trials have shown equivalent efficacy for the EGFR TKIs as compared with chemotherapy after failing one or two prior regimens [5,6].

The recent advances in the knowledge of the molecular landscape underlying lung cancer and its biological consequences have profound implications for lung cancer treatment that has been oriented to a biomarker-tailored approach. Patients with tumors bearing EGFR activating mutations derive a much higher benefit from EGFR TKIs as compared with those with EGFR wild-type tumors, and therefore erlotinib and gefitinib are the most recommended treatment alternatives for them, regardless the line of therapy [2,7]. Similarly, anaplastic lymphoma kinase (ALK) inhibitors have shown superiority over chemotherapy in tumors with EML4-ALK rearrangements [8]. Other targeted therapies are being tested in specific tumors with genomic alterations that make more biologically plausible the activity of specific inhibitors (of PI3K, Ros, MEK, Braf, and so on).

The efficacy of EGFR inhibitors in tumors without activating EGFR mutations has been a controversial issue, particularly their relative efficacy over the available chemotherapy in the second- and third-line setting. The recently published Tailor trial claims the superiority of docetaxel over erlotinib in EGFR wild-type patients in terms of OS (HR: 0.73; p = 0.05) and progression-free survival (PFS; HR: 0.71; p = 0.02) [9]. The robustness of this study is somehow limited due to its size and the major amendments in trial design and endpoints implemented after an interim analysis; however, secondary analysis of small subsets of nonmutated EGFR patients in a number of other trials suggests that chemotherapy may result in prolonged PFS compared with EGFR TKIs [5,6,10]. The questions remaining are if some advanced NSCLC patients with EGFR wild-type may derive relevant benefit from EGFR TKIs, and which tools may be useful for their identification.

Development of a plasma/serum pretreatment proteomic test: VeriStrat

Taguchi et al. developed a serum proteomic test using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry methodology aiming at predicting benefit from EGFR treatment [11]. Mass spectra from 139 patient’s serum samples before gefitinib treatment were used to define 8 MS features (i.e., peaks), that differentiated two outcome groups (long-term stable disease or early progression) on gefitinib therapy on the training set. An algorithm utilizing the mentioned MS features and based on k-nearest neighbors classification scheme was created, and its parameters were further optimized using additional spectra from the training cohort. The currently available commercial test, VeriStrat, assigns each sample a classification of VeriStrat Good (VSG) or VeriStrat Poor (VSP) using the same fixed set of parameters established during the development of the test. For a small number of samples (1–3%), an unequivocal classification cannot be determined. In the initial single-arm validation cohorts of patients treated with gefitinib (67 patients) and erlotinib (96 patients), the favorable proteomic signature subsets enjoyed improved survival (HR of 0.50 and 0.41 for gefitinib and erlotinib, respectively). The proteomic classifier had no significant discriminative outcome ability in three cohorts that did not receive EGFR TKIs.

Interestingly, although the VeriStrat test was developed in serum samples from lung cancer patients, at the present time, there is evidence suggesting that many (if not all) malignant diseases, but not benign conditions, would have a variable proportion of patients with the VSP spectra in plasma. This signature appeared preferentially associated with poor outcome and in some cases may predict for the benefit from specific therapies, as is the case for endocrine-dependent breast cancer [12].

A number of peaks in the mass spectra seem to be specific isoforms of serum amyloid A (SAA), which is secreted during the acute phase of inflammation and was overexpressed in poor prognosis classified patients [13]. This might indicate that inflammatory background is different between those characterized patients VSG or VSP. In the HCC4006 human cell line, characterized with EGFR exon 19 deletion, it has been showed that there was a relative decline in inhibition of these cells when grown in the presence of serum of patients with poor outcomes [14]. These results may suggest that the relative efficacy of targeted treatment is, at least partially, due to the host–tumor interaction.

Prognostic & predictive value of VeriStrat

The assessment of the differential prognostic and predictive value of a given biomarker requires the analysis of its performance in the context of randomized clinical studies. The comparison of the outcome on the control arm (e.g., placebo or chemotherapy) between biomarker positive and negative patients provides information on its prognostic capability. The differential effect on outcome of the experimental treatment in biomarker positive and negative patients is suggestive of a predictive role. Accordingly, the VeriStrat algorithm has been interrogated retrospectively and prospectively in samples from randomized trials, in addition to those from several single-arm studies where VSG patients confirmed to have improved outcomes in erlotinib-based treatments [15,16].

The VeriStrat test was performed to 441 pretreatment plasma samples from 731 patients included in the BR21 trial; 436 patients (98.9%) were successfully classified as VSG (61%) or VSP (39%) [17]. Among the 144 placebo-treated patients, VeriStrat was prognostic, with good patients having a significantly longer survival (HR: 0.44; p < 0.00001) and PFS (HR: 0.59; p < 0.0016) than poor patients. Those results were independent of other clinical variables and confirmed on a multivariate analysis. When the treatment arms were compared according to the VeriStrat status, the outcome was more favorable for erlotinib in VSG patients (median survival: 10.5 vs 6.6 months; HR: 0.63; p = 0.002) and VSP patients (median survival: 4 vs 3.1 months; HR: 0.77; p = 0.11), and the magnitude of the benefit was not dependent on the VeriStrat signature (interaction test p = 0.48). Similar results were obtained for PFS, indicating that the proteomic test was not predictive of outcome, except for response (p = 0.002).

The recently reported PROSE trial has provided more convincing evidence on the true prognostic and predictive value of VeriStrat [18]. Patients candidates (n = 285) for second-line treatment were classified as VSG or VSP and then within each stratum randomized to receive erlotinib or chemotherapy (docetaxel or pemetrexed). Overall, survival of the 263 patients evaluable for the primary analysis was similar in the two study arms (HR: 0.88; p = 0.313). Among the 128 patients treated on the control arm, survival was significantly longer for those VSG (median OS: 10.9 months) compared to those VSP (median OS: 6.4 months) (HR: 0.51; p = 0.0008), confirming the prognostic information associated with the signature. In addition, the test showed to be predictive of erlotinib efficacy (survival), as only VSG patients benefited from such a treatment. Among the 184 patients with favorable signature (70%), survival was comparable for erlotinib and chemotherapy (median survival: 10.95 vs 10.92 months; HR: 0.94; p = 0.714). In contrast, survival of the 79 VSP patients was superior for those treated with chemotherapy (median survival: 6.38 months) compared with those that received erlotinib (median survival: 2.98 months) (HR: 0.58; p = 0.022). Accordingly, the treatment–VeriStrat interaction test was significant despite adjustment for other covariates (HR: 1.98; p = 0.022) in the whole population and in the subset with tumors EGFR wild-type or unknown (HR: 1.94; p = 0.024). Response and PFS analysis from this study are awaited in order to confirm the predictive performance of VeriStrat in relation to these more immediate endpoints although independent imaging review was not implemented. Complementary information to that coming from the Prose trial, that only included a small proportion of patients with squamous cell histology, will result from the EMPHASIS Phase III study that will include 500 patients affected with this tumor subtype, and patients will be randomized to docetaxel or erlotinib treatment after stratification based on the VeriStrat classifier and performance status. Although the performance of VeriStrat seems independent from the EGFR status, it would be very interesting to review the outcome of the few cases described with EGFR mutation and VSP serum signature [19].

What is the most likely use of VeriStrat in the clinical practice? The currently available preliminary data suggest that VeriStrat may be clinically useful test to guide treatment decisions for NSCLC patients in the second-line setting. The main advantage of the information provided by the proteomic test would be the avoidance of treatment with erlotinib in the one-third of patients with VSP serum signature to whom chemotherapy is likely to result in improved outcomes. The test results may position VRG patients as appropriate candidates for both chemotherapy and erlotinib, and then treatment decisions are likely not to be driven by efficacy but toxicity, convenience and cost. Studies confined that the predictive capacity of VeriStrat for EGFR TKIs following two or more lines of chemotherapy would be also of interest, so that only patients likely to extract some benefit would be candidates for such a treatment.

Conclusions

The pretreatment plasma or serum proteomic classifier VeriStrat is a commercially available tool that has confirmed prognostic value in patients with NSCLC, who have failed to at least one line of prior chemotherapy and in other settings such as advanced endocrine-dependent breast cancer. In addition, the veristrat algorithm seems to specifically predict the benefit from EGFR TKIs, so that only patients with the favorable signature may benefit from such a treatment. Further studies should confirm and further define its role in predicting EGFR TKI benefit, particularly in squamous cell carcinoma or adenocarcinoma EGFR wild-type patients in the presence of KRAS mutation. In addition, the test has to establish its better use in terms of clinical efficiency identifying which patients are candidates for the test and at which time on the history of the disease, and lastly at which extra cost [20].

Financial & competing interests disclosure

L Paz-Ares is funded by Fondo de Investigacion Sanitaria (PI1102688) and RTICC (R12/0036/0028). MD Pastor is funded by Fondo de Investigacion Sanitaria (CD0900148.). S Molina-Pinelo is funded by Fondo de Investigación Sanitaria (CD1100153), Fundación Científica de la Asociación Española Contra el Cáncer, Consejería de Salud y Bienestar Social PI-0046-2012, and Fundación Mutua Madrileña (2010). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.