Stem-like profile is associated with poor prognosis in patients with low- or mid-range Gleason scores.
Scientists have found that classifying prostate tumors according to their mRNA expression patterns can help predict disease aggressiveness and poor prognosis in patients with low- or mid-range Gleason scores. A team led by researchers at the Institute for Advanced Study’s Simons Center for Systems Biology in Princeton analyzed mRNA microarray data from two separate cohorts of prostate cancer patients, and found that tumors manifesting stem-like signatures together with p53 and PTEN inactivation were indicative of very poor survival outcome.
The classification system, which included a number of different expression profile groupings, was independent of Gleason score. Arnold J. Levine, M.D., and colleagues describe their results in PNAS in a paper titled “Molecular classification of prostate cancer using curated expression signatures.”
The best current indicator of prostate cancer prognosis is a high Gleason score, which is determined on the pathology of the tumor section, the researchers explain. However, a significant number of patients with low Gleason scores do go on to develop aggressive disease and have poor survival.
For other cancers, such as breast cancer, meanwhile, microarray analysis of mRNA can be used to help define subclasses of disease, and embryonic stem cell (ESC) gene-expression profiles in breast cancers have been shown to correlate with p53 mutations and with subclasses of breast cancer associated with poorer outcome. Separate work, meanwhile, has defined for induced pluripotent stem cell (iPSC) expression signatures, and the polycomb repressive complex-2 (PRC2).
Given that such expression signatures, and in particular the ESC signature, can help provide prognostic information for breast cancers, the researchers postulated that it should also be possible to apply this same approach to other cancer types, and in particular prostate cancers.
To evaluate this further they looked at a recently published microarray dataset characterizing 281 prostate cancers from a watchful-waiting cohort recruited in Sweden between 1977 and 1999, together with associated clinical information including age, Gleason scores, cancer-specific mortality, and survival times. The tumors were tested for a set of different transcriptional signature profiles published in the literature to represent stemness features, and unsupervised clustering analysis. The same profiling procedure was then carried out on 150 prostate tumor samples from an independent dataset from patients at Memorial Sloan-Kettering Cancer Center.
Based on their previous work indicating that an ESC signature was closely associated with both p53 mutations and a very poor prognostic outcome, the team’s initial analyses focused on whether any of the Swedish cohort tumors expressed an ESC, iPSC, or PRC2 signature. The ESC signature was identified in 13% of the tumors, and a large proportion of these also had high Gleason scores. The iPSC signature, which was present in about 30% of the tumors, was found across the range of Gleason scores. This finding differed from prior evaluations of breast cancer, which indicated that ESC and iPSC signatures were more closely correlated, the authors note. In contrast with the ESC signature, the PRC2 signature was observed in 44% of the tumors and was more likely to be found in those with a low Gleason score.
Moreover, when tumors were clustered into separate ESC, PRC2, and indeterminate (i.e., no significant enrichment for either signature) groups, the average Gleason score was significantly lower in PRC2 tumors than in ESC tumors, and the ESC tumors had poorer survival.
The next stage was to investigate the molecular pathways active within the groups, so the researchers stratified the 281 prostate tumors using a library of diverse gene-expression signatures derived from a number of microarray datasets. “In particular, we formulated signatures indicating functionality of p53 and PTEN, presence of the TMPRSS2–ERG gene fusion, proliferation and MYC target activation, RAS pathway activity, and inflammatory signals,” the team notes. Unsupervised cluster analysis resulted in the generation of five clusters of prostate cancer, based on the signatures that were most significantly enriched.
These were: ESC | p53− | PTEN− (also referred to as stem-like), TMPRSS2–ERG fusion, cytokine | RAS | mesenchyme (also referred to as inflammatory-like), transitional, and PRC2 (also referred to as differentiated-like). The stem-like group (about 11% of the cohort) was associated with the poorest survival, and contained the vast majority of lethal cases. The TMPRSS2–ERG fusion signature group (another 18% of cases) was associated with the next-lowest survival outcome. The remaining three groups had statistically similar survival outcomes.
In fact, patients in the stem-like and fusion groups carried 3.3-fold higher, and 1.8-fold higher risks of mortality than patients in the other three groups. Moreover, of the 200 patients with Gleason scores of 6 and 7, the patients with a stem-like profile carried a 2.7-fold higher risk of dying of the disease, and the increase was 1.9-fold in patients with the fusion profile. “Together, the ESC | P53− | PTEN−group and the TMPRSS2–ERG group account for 40% of all prostate cancer deaths in this study,” the authors write. “Indeed the ESC | P53− | PTEN− group and the subgroup with a high Gleason score both had very similar predictive power for survival.”
Samples in the Swedish cohort were obtained using a clinically “untypical” method that involved transurethral resection of the prostate, paraffin embedding of tissue, and the measurement of just 6,144 genes on an Illumina custom chip. The team therefore repeated their signature profile analysis on the more clinically relevant samples from the second Memorial Sloan-Kettering Cancer Center cohort, which were obtained at radical prostatectomy, fresh-frozen, and analyzed on an Affymetrix Human Exon microarray. Compared with the Swedish study, the Sloan-Kettering dataset displayed different statistical parameters, including a lower average Gleason score, a far lower proportion of high-grade tumors, and lower average age at diagnosis. In addition, survival time was recorded but heavily censored, with only one death from prostate cancer occurring within the five-year study period.
The Sloan-Kettering signature scores were calculated using gene-set enrichment analysis and unsupervised clustering carried out, as with the Swedish dataset, and samples stratified into four stable groups that had very similar association patterns to the Swedish set. This resulted in a group classified as ESC | (P53−) | PTEN−, which comprised stem-like tumors analogous to the stem-like group in the first set. In this group there was a weak signal for loss of p53 function, but this was not significant, possibly due to the small signature size, or the earlier age at diagnosis (as p53 mutations are late events in prostate cancer), the authors note. There was also a TMPRSS2–ERG fusion group, a differentiated PRC2 group, and one cytokine | transitional group, which was comparable to the combined cytokine | RAS | mesenchyme, and transitional subtypes within the Swedish dataset.
As with the previous analyses, the stem-like group harbored the most aggressive cancers, including 70% of the metastatic tumor samples, and was associated with significantly shorter time to PSA recurrence and high Gleason score. The fusion group was intermediate, while the cytokine | transitional group was benign. The differentiated group (PRC2) included 9% of the metastatic samples, and included 20% of metastatic events, which was comparable to the fusion group, the researchers note. Overall, the ESC | (P53−) | PTEN− group contained 17% of all cases with recurrent or metastatic disease and 33% in the fusion group; effectively 50% of all such cases recorded in the five-year follow-up period.
“Considering the extensive differences between the two cohorts, we observed a remarkable correspondence between the molecular profiles and their associated outcomes in both groups,” the authors state. “This work is a unique classification of prostate tumors into subgroups with distinct survival outcomes based upon microarray data. The analysis was able to detect two structurally different groups of patients carrying increased risk, with the stem-like tumors being the most aggressive subtype, in two fundamentally different datasets and across different sampling techniques. “This classification is independent of Gleason score and therefore provides useful unique molecular profiles for prostate cancer prognosis, helping to predict poor outcome in patients with low or average Gleason scores.”