Improved prediction of prostate cancer recurrence through systems pathology.

We have developed an integrated, multidisciplinary methodology, termed systems pathology, to generate highly accurate predictive tools for complex diseases, using prostate cancer for the prototype. To predict the recurrence of prostate cancer following radical prostatectomy, defined by rising serum prostate-specific antigen (PSA), we used machine learning to develop a model based on clinicopathologic variables, histologic tumor characteristics, and cell type-specific quantification of biomarkers. The initial study was based on a cohort of 323 patients and identified that high levels of the androgen receptor, as detected by immunohistochemistry, were associated with a reduced time to PSA recurrence. The model predicted recurrence with high accuracy, as indicated by a concordance index in the validation set of 0.82, sensitivity of 96%, and specificity of 72%. We extended this approach, employing quantitative multiplex immunofluorescence, on an expanded cohort of 682 patients. The model again predicted PSA recurrence with high accuracy, concordance index being 0.77, sensitivity of 77% and specificity of 72%. The androgen receptor was selected, along with 5 clinicopathologic features (seminal vesicle invasion, biopsy Gleason score, extracapsular extension, preoperative PSA, and dominant prostatectomy Gleason grade) as well as 2 histologic features (texture of epithelial nuclei and cytoplasm in tumor only regions). This robust platform has broad applications in patient diagnosis, treatment management, and prognostication.

Personalized prediction of tumor response and cancer progression on prostate needle biopsy.

PURPOSE: To our knowledge in patients with prostate cancer there are no available tests except clinical variables to determine the likelihood of disease progression. We developed a patient specific, biology driven tool to predict outcome at diagnosis. We also investigated whether biopsy androgen receptor levels predict a durable response to therapy after secondary treatment. MATERIALS AND METHODS: We evaluated paraffin embedded prostate needle biopsy tissue from 1,027 patients with cT1c-T3 prostate cancer treated with surgery and followed a median of 8 years. Machine learning was done to integrate clinical data with biopsy quantitative biometric features. Multivariate models were constructed to predict disease progression with the C index to estimate performance. RESULTS: In a training set of 686 patients (total of 87 progression events) 3 clinical and 3 biopsy tissue characteristics were identified to predict clinical progression within 8 years after prostatectomy with 78% sensitivity, 69% specificity, a C index of 0.74 and a HR of 5.12. Validation in an independent cohort of 341 patients (total of 44 progression events) yielded 76% sensitivity, 64% specificity, a C index of 0.73 and a HR of 3.47. Increased androgen receptor in tumor cells in the biopsy highly significantly predicted resistance to therapy, ie androgen ablation with or without salvage radiotherapy, and clinical failure (p <0.0001). CONCLUSIONS: Morphometry reliably classifies Gleason pattern 3 tumors. When combined with biomarker data, it adds to the hematoxylin and eosin analysis, and prostate specific antigen values currently used to assess outcome at diagnosis. Biopsy androgen receptor levels predict the likelihood of a response to therapy after recurrence and may guide future treatment decisions.

Multifeature prostate cancer diagnosis and Gleason grading of histological images.

We present a study of image features for cancer diagnosis and Gleason grading of the histological images of prostate. In diagnosis, the tissue image is classified into the tumor and nontumor classes. In Gleason grading, which characterizes tumor aggressiveness, the image is classified as containing a low- or high-grade tumor. The image sets used in this paper consisted of 367 and 268 color images for the diagnosis and Gleason grading problems, respectively, and were captured from representative areas of hematoxylin and eosin-stained tissue retrieved from tissue microarray cores or whole sections. The primary contribution of this paper is to aggregate color, texture, and morphometric cues at the global and histological object levels for classification. Features representing different visual cues were combined in a supervised learning framework. We compared the performance of Gaussian, k-nearest neighbor, and support vector machine classifiers together with the sequential forward feature selection algorithm. On diagnosis, using a five-fold cross-validation estimate, an accuracy of 96.7% was obtained. On Gleason grading, the achieved accuracy of classification into low- and high-grade classes was 81.0%.

A systems pathology model for predicting overall survival in patients with refractory, advanced non-small-cell lung cancer treated with gefitinib.

PURPOSE: To identify clinical and biometric features associated with overall survival of patients with advanced refractory non-small-cell lung cancer (NSCLC) treated with gefitinib. EXPERIMENTAL DESIGN: One hundred and nine diagnostic NSCLC samples were analysed for EGFR mutation status, EGFR immunohistochemistry, histologic morphometry and quantitative immunofluorescence of 15 markers. Support vector regression modelling using the concordance index was employed to predict overall survival. RESULTS: Tumours from 4 of 87 patients (5%) contained EGFR tyrosine kinase domain mutations. A multivariate model identified ECOG performance status, and tumour morphometry, along with cyclin D1, caspase-3 activated, and phosphorylated KDR to be associated with overall survival, concordance index of 0.74 (hazard ratio (HR) 5.26, p-value 0.0002). CONCLUSIONS: System-based models can be used to identify a set of baseline features that are associated with reduced overall survival in patients with NSCLC treated with gefitinib. This is a preliminary study, and further analyses are required to validate the model in a randomised, controlled treatment setting.

Systems pathology approach for the prediction of prostate cancer progression after radical prostatectomy.

PURPOSE: For patients with prostate cancer treated by radical prostatectomy, no current personalized tools predict clinical failure (CF; metastasis and/or androgen-independent disease). We developed such a tool through integration of clinicopathologic data with image analysis and quantitative immunofluorescence of prostate cancer tissue. PATIENTS AND METHODS: A prospectively designed algorithm was applied retrospectively to a cohort of 758 patients with clinically localized or locally advanced prostate cancer. A model predicting distant metastasis and/or androgen-independent recurrence was derived from features selected through supervised multivariate learning. Performance of the model was estimated using the concordance index (CI). RESULTS: We developed a predictive model using a training set of 373 patients with 33 CF events. The model includes androgen receptor (AR) levels, dominant prostatectomy Gleason grade, lymph node involvement, and three quantitative characteristics from hematoxylin and eosin staining of prostate tissue. The model had a CI of 0.92, sensitivity of 90%, and specificity of 91% for predicting CF within 5 years after prostatectomy. Model validation on an independent cohort of 385 patients with 29 CF events yielded a CI of 0.84, sensitivity of 84%, and specificity of 85%. High levels of AR predicted shorter time to castrate prostate-specific antigen increase after androgen deprivation therapy (ADT). CONCLUSION: The integration of clinicopathologic variables with imaging and biomarker data (systems pathology) resulted in a highly accurate tool for predicting CF within 5 years after prostatectomy. The data support a role for AR signaling in clinical progression and duration of response to ADT.