Prognostic Factors in Pleural Mesothelioma Patients Receiving First-Line Chemotherapy: Establishing the PLECH Baseline Risk Score

Pleural mesothelioma (PM) is a rare, yet highly aggressive cancer originating from the mesothelial lining of the pulmonary pleura, notorious for its poor prognosis. Although the precise global impact of mesothelioma remains uncertain, estimates suggest it may be responsible for as many as 38,400 deaths annually [1].

The current median overall survival (OS) rate for patients with PM typically ranges between 14 and 22 months [2, 3].The majority of patients are diagnosed in advanced stages, making resectable disease uncommon. Even when surgery is an option, it often leads to a significant increase in morbidity; additionally, trials have not been able to consistently demonstrate a survival benefit [4]. Consequently, the focus for the majority of patients shifts to systemic therapies, for which most patients receive platinum-based chemotherapy.

Within the context of poor outcomes, predicting prognosis has emerged as an important objective for oncologists. This is to better meet the needs of individual patient populations and to balance the adverse effects of chemotherapy against its potential survival benefits [5]. While individual factors like non-epithelioid histology and performance status (PS) are associated with outcomes, it has become increasingly clear that prognosis prediction in mesothelioma is complex. It is not reliant on single factors but is likely influenced by a combination of various clinical and para-clinical characteristics [6, 7].

Thus far, the primary tools for prognosis assessment in this field have traditionally been prognostic scores, with the European Organisation for Research and Treatment of Cancer (EORTC) and the Cancer and Leukemia Group B (CALGB) scores being the most prominent and validated [8]. However, these scoring systems face notable limitations. Derived, respectively, from five and seven phase II mesothelioma clinical trials in the 90s in Europe and North America [9, 10], there is a concern about their relevance to contemporary treatments and their applicability across different populations. While various cohorts have replicated these scores, supporting their validity [6, 11, 12], other studies have found these scores not to have a strong prognostic value [5, 13]. Notably, certain demographics, like Hispanic patients, have been underrepresented in these studies. Moreover, since these trials included surgical patients, their relevance is lessened for those with unresectable disease receiving the first-line chemotherapy [9, 10]. Another limitation is the complexity of calculating these scores, which has confined their use mostly to clinical trials, rather than in everyday clinical practice.

In efforts to fill this knowledge gap, new prognostic scores have been developed in recent years [14‒16]. However, they often face limitations, such as development from small population sizes due to the rarity of this cancer, the inclusion of costly markers, and insufficient representation of certain demographics, like Hispanics. In this context, it is vital to have dependable tools for prognosticating, especially for patients undergoing first-line chemotherapy, who constitute the majority and can benefit from enhanced treatment decision-making. Specifically, it is crucial to identify patients who are suitable candidates for chemotherapy as opposed to those for whom alternative strategies might be more appropriate [7].

The aim of the present study was to identify baseline, readily available, prognostic factors associated with OS in patients with PM to develop a simple prognostic score. This score will be based on data from a large multicenter cohort, encompassing two of the largest cancer centers in Mexico.

We conducted a retrospective cohort study at Mexico’s National Cancer Institute and Mexico’s National Medical Center Century XXI. The study identified patients with unresectable PM who received the first-line chemotherapy between 2010 and 2023.

We included patients with histologically confirmed unresectable PM, aged ≥18 years at diagnosis, who received the first-line chemotherapy. We excluded patients with a history of surgical resection and those lacking baseline laboratory data at diagnosis (hemoglobin, albumin, lymphocytes, neutrophils, platelets, LDH). Electronic medical records were accessed to extract clinical and pathological variables, as well as baseline laboratory values before the initiation of chemotherapy. Patients without complete information were excluded.

All variables were included in an exploratory univariate survival analysis. These variables were preselected based on previously published data indicating their potential prognostic value and routine baseline availability. Variables showing some prognostic value in univariate the Kaplan-Meier analysis (p < 0.1) were included in a multivariate Cox proportional hazards regression analysis to identify independent prognostic factors significantly associated with OS. Following the score development methodology outlined by Zhang et al. [17], statistically significant variables were then analyzed independently to determine their regression coefficients, reflecting their independent prognostic contribution. A prognostic score was developed, assigning points to each variable based on the magnitude of its regression coefficient. To normalize these coefficients, we divided each one by the smallest coefficient and rounded to the nearest integer. This process makes all coefficients easier to compare and interpret [18, 19]. We calculated the score for the entire cohort, plotted ROC curves, and determined the area under the curve (AUC) to assess the score’s ability to predict 1-year OS. The optimal cut-off point was determined using Youden’s J statistic, and the score was dichotomized into high versus low risk for evaluation using univariate the Kaplan-Meier analysis. Additionally, ROC curves and AUC were calculated for the CALGB and EORTC scores to compare their prognostic capacities with our developed index.

Categorical variables were reported as frequencies and proportions. Continuous variables were dichotomized to allow the inclusion of all variables in conventional cancer survival analysis (Kaplan-Meier). These dichotomized variables were then likewise reported as frequencies and proportions. The dichotomization was based on optimal cut-off points derived from ROC curve analysis and Youden’s J statistic, as no widely validated cut-offs have been published for our patient population [20‒22]. The specific cut-offs used are provided in Table 1.

Our primary endpoint was OS, defined as the time from diagnosis to death. Our secondary endpoint was progression-free survival (PFS), defined as the time from initiation of chemotherapy to disease progression per Response Evaluation Criteria in Solid Tumors (RECIST) for PM Version 1.1 [23]. The Kaplan-Meier analysis was conducted for each variable based on dichotomized values, and both median OS and PFS with 95% confidence intervals were reported. Group comparisons were made using the log-rank test.

Analyses were performed using SPSS software (version 26.0; IBM, Chicago, IL, USA) and ROC curves with optimal cut-off points were calculated using the “p-roc” package in RStudio (Version 2023.09.1+494). Graphs were plotted using GraphPad Software (Prism version 9.5.0, San Diego, CA, USA).

We identified 335 patients diagnosed with mesothelioma. Of these, 262 patients with unresectable mesothelioma were treated with the first-line chemotherapy and complete clinical information in the electronic medical record met our inclusion criteria. The majority were male (69.1%) with a good functional status (ECOG <2 in 78.2% of patients). Epithelioid histology was the most common (80.5%). Detailed baseline characteristics are presented in Table 1.

The median OS for the entire cohort was 16.2 months (95% CI 13.8–18.5), and the median PFS was 7.8 months (95% CI 6.7–8.8). Most patients received a platinum-based regimen: 155 coupled with gemcitabine, 67 with pemetrexed, 10 with doxorubicin, and 8 in monotherapy. Additionally, 3 patients received pemetrexed monotherapy, 3 vinorelbine monotherapy, and 16 participated in research protocols involving chemotherapy. No significant differences were observed in OS and PFS across types of treatment (p = 0.245 and p = 0.636, respectively).

Univariate analysis revealed that asbestos exposure, non-epithelioid histology, advanced clinical stage, poor PS (ECOG ≥2), chest pain at diagnosis, weight loss, elevated levels of LDH, platelet count, neutrophil count, leukocyte count, and low levels of albumin and hemoglobin were all associated with worse OS (Table 1). For PFS, factors such as non-epithelioid histology, advanced clinical stage, poor PS, chest pain at diagnosis, high platelet count, neutrophil count, leukocyte count, and low levels of albumin and hemoglobin were statistically significant (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000543637).

In the OS multivariate analysis, non-epithelioid histology, poor PS (ECOG ≥2), chest pain at diagnosis, high LDH, and high platelet count remained significantly associated with a poor prognosis (online suppl. Table 2). After pooling these results and normalizing individual regression coefficients, the following scores were assigned: high platelet count (+2), high LDH (+1), ECOG ≥2 (+1), chest pain at diagnosis (+2) and non-epithelioid histology (+1). These values constitute the basis of what we have termed the PLECH risk score, which has a possible range of 0–7 points (Table 2). The ROC curve for predicting 1-year OS showed an AUC of 0.70, with an optimal cut-off point at 2.5, yielding a sensitivity of 75% and a specificity of 55% (online suppl. Fig. 1). ROC curves for calculated CALGB and EORTC scores in our population showed AUCs of 0.60 and 0.57, respectively (Fig. 1). In the Kaplan-Meier analysis, patients with a high-risk score (≥3) demonstrated significantly worse OS (12.3 months [95% CI 9.9–14.6] vs. 20.1 months [95% CI 14.1–25.9]; p < 0.001) and PFS (6.4 months [95% CI 5.3–7.5] vs. 11.3 months [95% CI 8.3–14.3]; p < 0.001) (Fig. 2).

To further evaluate the prognostic significance of the PLECH score, we conducted an independent analysis in our two included centers. In center 1 (Mexico’s National Cancer Institute), patients with a high-risk score (≥3) exhibited significantly poorer OS compared to those with lower scores, with a median OS of 11.7 months (95% CI 9.0–14.4) versus 17.0 months (95% CI 10.2–23.8), respectively (p < 0.001). The AUC for the PLECH score in this center was 0.68.

In center 2 (Mexico’s National Medical Center Century XXI), the results were consistent, with high-risk patients (score ≥3) showing markedly reduced OS, with a median OS of 12.5 months (95% CI 8.8–16.2) compared to 22.2 months (95% CI 15.6–28.8) for those with lower scores (p < 0.001). The AUC in this center was 0.70, further supporting the score’s prognostic value.

Given that histology and PS are universally acknowledged prognostic markers in PM, we subsequently developed a 3-variable score incorporating only the additional PLECH variables (LDH, platelets, chest pain), which yielded an AUC of 0.65. We also assessed a 2-variable score using only histology and PS, with an AUC of 0.60. Both scores were lower than the full 5-variable PLECH score, which provided the highest prognostic accuracy (AUC = 0.70), underscoring the combined value of all variables.

This cohort delineates epidemiological characteristics of mesothelioma patients receiving the first-line chemotherapy, with a predominance in males (∼70%) and a majority exhibiting epithelioid histology (∼80%), consistent with global PM data [24‒26]. The median diagnosis age was 64 years, consistent with Asian and Latin-American cohorts [14, 26], yet younger than the 70–75 years observed in UK and US populations [27, 28]. Notably, our population shows a lower documented asbestos exposure rate (∼40%) compared to the widely recognized 85% of PM cases in the literature [27]. As asbestos exposure in Mexican patients is not negligible, this discrepancy likely reflects underdiagnosed or unrecognized exposure [29]. Similar patterns in other cohorts suggest potential underestimation or lack of awareness [30].

Multivariate analysis identified five factors independently associated with poor OS (Table 2), included in our prognostic index. Epithelioid histology, a universally acknowledged prognostic marker in mesothelioma, significantly improved OS compared to sarcomatoid and biphasic types [25]. The underlying mechanisms for these enhanced outcomes in patients with epithelioid histology remain somewhat elusive. However, potential mechanisms include distinct biomarker profiles, such as lower PD-L1 expression and BAP1 loss, which correlates with improved outcomes in platinum/pemetrexed-treated patients [31, 32]. Additionally, the expression of mesothelin, prevalent in epithelioid cases, may influence prognosis [33].

PS, an additional classically recognized prognostic factor, particularly for chemotherapy recipients, showed an independent association with poor OS. Poor PS impacts chemotherapy tolerance, and current guidelines suggest it should be considered when deciding on alternative treatments, such as using single-agent chemotherapy or palliative care alone [34‒36]. Accordingly, chest pain at diagnosis, indicative of advanced disease or complications like pleural effusion and associated with poorer quality of life, was also prognostic of poor OS [37, 38].

Elevated LDH, an intracellular glycolytic enzyme, overexpressed in cancer cells and an accessible serum laboratory marker linked to increased cellular turnover, was a significant predictor of poor OS, but not PFS [39]. In the context of mesothelioma, the association of LDH with adverse outcomes has been observed, though findings across studies are not entirely consistent. A meta-analysis, encompassing data from 1,977 patients across both retrospective and prospective cohorts, revealed a significant correlation between high pretreatment LDH levels and reduced OS [40]. Similarly, high platelet count, known to support tumor growth and immune evasion, was associated with worse OS. These findings are consistent with broader cancer studies and meta-analyses, establishing platelets as prognostic markers [41‒43].

The association between high LDH and platelet counts with adverse outcomes was evident when utilizing the optimal cut-off values derived from our population data, of 292.5 U/L and 318 × 10⁹/L, respectively. Notably, our LDH cut-off aligns closely with thresholds commonly recognized by clinical laboratories, suggesting biological relevance [44]. Conversely, while the platelet count cut-off is marginally lower than the conventional upper limit of normal of 400 × 10⁹/L, it is consistent with the cut-off points of 300–400 × 10⁹/L frequently employed in mesothelioma studies [45]. Information across different cancers has shown that even patients with elevations approaching the upper limit of normal have worse outcomes, indicating that “normal” ranges can be context-dependent [42]. The origin of these specific cut-offs, whether reflective of intrinsic population variability, including ethnic differences, or due to inter-sample variability, remains unclear [46].

Altogether, the conjoined predictive power of all the variables independently shown to have prognostic value in our study, including readily obtained measures from a large cohort receiving the first-line chemotherapy, led to the development of the PLECH score, which showed to be better at 1-year OS prediction compared to CALGB and EORTC, as detailed in Figure 1. This study does validate several prognostic factors from the CALGB and EORTC, namely, PS and chest pain. However, it does so within the confines of a more uniform cohort – specifically, patients with unresectable mesothelioma who have received the first-line chemotherapy, as opposed to the diverse groups that constituted the CALGB and EORTC trials. Moreover, the PLECH score is tailored to modern chemotherapy regimens, including cisplatin/pemetrexed and cisplatin/gemcitabine [47, 48], diverging from the varied treatments used in the CALGB and EORTC trials. These findings suggest a potentially reduced benefit of chemotherapy in high-risk patients, questioning whether these individuals might derive greater benefit from alternative approaches, such as immunotherapy, which was not explored in this study but should be assessed in independent cohorts.

Naturally, this study is not without limitations. Its retrospective nature and the fact that it is based on data from only two centers in a single Latin-American country may limit the generalizability of the findings. The development of the score is constrained by the unavailability of novel biomarkers within our setting and the absence of certain baseline laboratory parameters that are not routinely ordered at our centers. Furthermore, the dichotomization of continuous variables using ROC curve analysis, while tailored to the characteristics of our population, may have constrained the prognostic capacity of the included variables when compared to alternative dichotomization methods. Additionally, the findings have not yet undergone validation in an independent cohort, underscoring the need for further research.

Despite these limitations, the PLECH score stands out for its practicality, requiring purely clinical information, which is routinely obtained, in order to calculate a simple 7-point score. Its implementation could be particularly beneficial in settings where advanced biomarkers are not readily available. Moving forward, there remains a clear need to identify and integrate novel biomarkers that have been validated in Hispanic populations with clinical scores such as this one, to enhance prognostic capabilities and guide treatment more effectively.

This study enhances our understanding of prognostic factors in PM within a significant cohort, underscoring the importance of non-epithelioid histology, poor PS, chest pain, high LDH, and elevated platelet count. By introducing the PLECH score as a potential prognostic tool, which demonstrated superior predictive capacity compared to traditional CALGB and EORTC scores, it further underscores the necessity for tailored approaches in PM management. These findings call for additional validation and research to refine prognostic assessments, including the integration of biomarkers and the development of treatment strategies, with the goal of enhancing patient outcomes across diverse settings, especially for those within high-risk groups.

This study protocol was reviewed and approved by the Comité de Investigación of the Instituto Nacional de Cancerología, approval No. (2023/007). The need for informed consent was waived by the aforementioned committee.

Dr. Oscar Arrieta reports personal fees from Pfizer, Lilly, Merck, and Bristol Myers Squibb, grants and personal fees from Astra Zeneca, Boehringer Ingelheim, and Roche.

This research project did not receive any external funding.

A.G. contributed to conceptualization, methodology, formal analysis, investigation, data curation, and writing the original draft, as well as review and editing. L.A.C.-M. contributed to methodology, investigation, writing review and editing, supervision, and project administration. A.P.G.-G. contributed to methodology, formal analysis, investigation, and writing the original draft and review and editing. R.T.-R. contributed to methodology, data curation, and investigation. W.M.-M. contributed to methodology, data curation, and investigation. D.F. contributed to investigation, writing review and editing, and supervision. N.R. contributed to methodology, formal analysis, investigation, and writing review and editing. O.A. contributed to conceptualization, methodology, investigation, resources, writing review and editing, supervision, and project administration.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants, but they are available from the corresponding author O.A. upon reasonable request.