Evaluation of clonal origin of malignant mesothelioma

Malignant mesothelioma (MM) is an aggressive cancer arising from the transformation of the mesothelial lining of the pleura, peritoneum and pericardium. It is a lethal cancer affecting approximately 3,200 individuals each year in the US, most of them die within 1 year from diagnosis [1]. MM incidence has remained stable in the US since 2003, but it continues to increase worldwide, due to exposure to asbestos fibers, which are widely used for industrial purposes [1]. Occupational exposure accounts for a male to female MM incidence ratio of 6-8 to 1, as males are much more often involved in the asbestos industrial trades [2]. Asbestos, erionite and other mineral fibers are also naturally present in developing rural areas where they pose a major risk factor for MM [3],[4]. Asbestos inhalation induces a chronic inflammatory response at sites of fibers deposition that may lead to malignant mesothelial cell transformation after a latency of 30- to 50-years [5]. In addition to the environmental and inflammatory components of MM etiology, we have recently discovered that germline mutations of the BAP1 gene cause a novel cancer syndrome characterized by a very high incidence of MM, and other malignancies [6].

MMs show different histology. There are 3 main subtypes: about 50% of MM show an epithelioid morphology and they look like carcinomas, 10% have a spindle cell morphology similar to sarcomas, and about 35% are biphasic, composed of both epithelioid and spindle cells in different proportions. In addition there are less common histological variants [7]. Diagnosis is further complicated by the presence of intra-tumoral pleomorphisms and phenotypic heterogeneity, raising the question of whether MM results from genetic and epigenetic alterations, which drive clonal tumor evolution into these different morphologies, or it arises from different subsets of mesothelial cells that interact or cooperate to drive malignant progression.

Multistep carcinogenesis is the currently accepted hypothesis to explain genetic diversity in tumors [8]-[10]. This hypothesis is based on the notion that somatic mutations are rare events that are unlikely to co-occur in different single normal cells, and that the mutations leading to the genetic diversification of the tumor architecture occur during the process of clonal expansion and selection. A cancer is considered monoclonal when all cells within the tumor can be traced back to a single progenitor/initiator cell. Instead, a polyclonal malignancy derives from the concomitant transformation of two or more different ancestor cells. Determining the clonal status of a cancer can be quite challenging, because of the inherent plasticity of the cancer genome that can acquire many somatic mutations during malignant cell growth. Analysis of the X chromosome inactivation pattern in female cancer biopsies, by measuring the methylation status of the polymorphic human androgen receptor (HUMARA) locus, is considered the most accurate method to assess clonality [11]. During early female embryogenesis, one of the two X chromosomes is randomly inactivated and the pattern of X-chromosome inactivation is stably transmitted from parent cell to the progeny (Lyonization) [12]. Therefore, the presence of the same inactivated X chromosome in all cancer cells has been interpreted as an indication of monoclonality. Although the current dogma is that cancers originate as monoclonal growths, some studies suggest that some cancers may arise as polyclonal [13],[14].

To the best of our knowledge, the clonal origin of MM has never been investigated. However MMs are “assumed” to be monoclonal as most tumors are usually assumed to be. Here we tested the hypothesis that MMs are monoclonal using the HUMARA assay on 16 MM biopsies from 14 female patients. We found that human MMs are polyclonal in origin.

To address whether MM have monoclonal or polyclonal origin, we performed the HUMARA assay, as described in Methods and Figure 1[11],[15],[17]. We first performed the HUMARA assay on control samples (Figure 2). A healthy male DNA sample (monoallelic, bearing a single unmethylated X chromosome) was digested as described in the material and methods and analyzed by agarose gel and capillary electrophoresis. Disappearance of the PCR band/peak indicated complete allele digestion, thus excluding a bias in our HUMARA assay due to incomplete HpaII digestion (Figure 2A). A healthy female DNA sample (L-IV-II, biallelic) and a known monoclonal melanoma cell line (#1290, female) were used as additional controls. As shown in Figure 2B, gel and capillary electrophoresis successfully detected a polyclonal and a monoclonal pattern, respectively.

Next, we established the sensitivity of the HUMARA assay in the detection of under-represented alleles by both gel and capillary electrophoresis (Figure 3). Mono-allelic (HpaII-digested) and bi-allelic (non-digested) DNA from the 1290 melanoma monoclonal cell line were mixed in different proportions. PCR products were resolved by electrophoresis on 3% agarose gel and visualized under UV light in the presence of ethidium bromide (Figure 3A), or by capillary electrophoresis using a Genotypic Bioanalyzer (Applied Biosystems) (Figure 3B). Linear regression of input vs detected allele ratios revealed a robust correlation (R² > 0.98), with the less frequent allele detectable when present at a fraction greater than or equal to 1/8 (12.5% of the input copies) (Figure 3C).

HUMARA clonality assessment was conducted on 16, stage I to III, frozen biopsies from 14 cases of female MMs (Table 1). In 2 out of 14 sporadic cases, two distinct nodules from the same pleura were available for analysis, giving a total of 16 specimens. Samples were classified as non-informative when a single band or peak was detected in the nearby normal tissue after digestion with the HpaII enzyme, indicating the presence of skewed Lyonization. Out of the 16 biopsies tested, 15 were informative. Sample L-III-18, a case of familial MM, was non-informative as PCR amplification of both mock- and HpaII-digested DNA produced a single band, suggesting that the lengths of the paternal and maternal alleles were identical (data not shown). Of the 15 informative biopsies, 14/15 PCR products (93.3%) displayed two distinct bands and peaks. 4 representative samples are shown in Figure 4A and B. Corrected Allele Ratio (CR) calculated on the allele peak areas by Genotypic Bioanalyzer was within the values of 0.33 and 3.0, indicating polyclonality of all 14 samples (Figure 4B and Table 1). The observed proportion of polyclonality was 93.33% (14/15), with a 95% confidence interval of [68.16% - 100%], indicating that the proportion of polyclonality in our sample population did not differ significantly from 100%. Therefore our results reveal with confidence that MM are polyclonal at origin. Case #524 showed a quite distinct pattern, as one nodule (#524B) revealed a monoclonal pattern (CR = 0.19), while the other (#524A) was polyclonal (CR = 0.36; Figure 4C). This finding suggested that a particular clone may dominate a certain area within a largely polyclonal tumor, composed of clones derived from different cells of origin. Of note, both sporadic MMs, that usually develop on a background of asbestos exposure, and familial MMs that develop on a background of germline BAP1 mutations were found to be polyclonal. All these MM samples were from women, as the HUMARA assay used here to assess clonality, cannot be used on specimen from men. MMs in women are rare and have a slightly better prognosis than MM in man [20], a finding that some studies suggested might be related to the expression of the estrogen receptor beta [21]. However, there is no evidence in the literature supporting the hypothesis that women and man MM have a different pathogenesis, thus it seems unlikely that polyclonality of MM is influenced by gender-associated factors. It should also be noted that all these MMs were either of the epithelial or of the biphasic type (Table 1), which account for 90% of MMs. We did not test sarcomatoid MMs that account for about 10% of all MMs, because these specimens were not available, as they are rare and as these patients are not candidates for surgical resection [2]. Thus, clonality of sarcomatoid MMs remains to be tested.

X-chromosome inactivation based assays may detect a seemingly monoclonal tumor when transformation occurs in multiple cells having the same inactivated X chromosome. Although possible, this is a rare event, mainly dependent on the X-inactivation patch size during early developmental Lyonization. This event leads few of the progeny of a single embryonic stem cell to be grouped together in the adult, forming patches. This phenomenon has been described in breast tissues, which display a rather large patch size [12]. Alternatively, a true monoclonal nodule may result from the clonal outgrow of a particular cell with proliferative advantage, as previously described in breast and prostate cancer, e.g. due to acquisition of LOH of the X-linked tumor suppressor FOXP3[22],[23]. The notion that tumors derive from a single cell through the expansion and evolution of several clones has survived mostly unchallenged until the present time. Accordingly, much effort has been placed in using genomic approaches to dissect the clonal relationships present within single tumors [8]. Despite the advances of the molecular genetic analyses, to date, the HUMARA assay is unique in its capability of detecting the actual origin of a cell population. Our data indicate that MMs may arise as polyclonal tumors due to concurrent transformation of multiple mesothelial cells (Figure 5). Recent studies conducted in some breast and colon carcinomas showed that they were polyclonal [13],[14],[24],[25], supporting the novel concept that not all tumors are monoclonal at origin. Also, a recent study using mouse chimeras, provides evidences suggesting that intestinal tumors may be polyclonal and that inter-clonal interactions are necessary for tumor development [13]. Our results that MMs are polyclonal indicate that MMs are likely to be clonally complex at the outset. The polyclonal origin of MM may account for the very high degree of intra-tumoral heterogeneity found in these malignancies, and also contribute to the emergence of drug-resistant subpopulations. Therefore, our finding suggest that solely tracking the clonal evolution of a predominant clone may not be sufficient to successfully target MM, providing a possible rationale to the peculiar resistance of MM to current therapies.

Our data indicate that MM arise as polyclonal tumors, a finding that has both pathogenesis and clinical implications. For example, MM patients whose tumors are removed at Stage Ia, most often experience recurrence after surgery in spite of apparent successful MM eradication [20].

Our data suggest that, in contrast to current dogma, recurrence may represent novel malignancies, occurring because of the carcinogenic “field effect” of asbestos, its related chronic inflammation, and/or because of ubiquitous genetic predisposition in patients carrying germline BAP1 mutations. Therefore, MM may arise from a large pool of independent and mostly covert cancers, as observed in some other malignancies [26]. Accordingly, the multiple minuscule pleural nodules that are characteristically found on the pleura of early-stage MM patients are likely pre-malignant lesions or early tumors rather than early local metastases. Our findings underscore the need to attack simultaneously several different molecular targets to try to eliminate the different MM cell clones, as each clone may carry its own distinct set of molecular alterations.