Gleason score as predictor of clinicopathologic findings and biochemical (PSA) progression following radical prostatectomy

Machine learning discrimination of Gleason scores below GG3 and above GG4 for HSPC patients diagnosis

This study aims to develop machine learning (ML)-assisted models for analyzing datasets related to Gleason scores in prostate cancer, conducting statistical analyses on the datasets, and identifying meaningful features. We retrospectively collected data from 717 hormone-sensitive prostate cancer (HSPC) patients at Yunnan Cancer Hospital. Of these, data from 526 patients were used for modeling. Seven auxiliary models were established using Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Extreme gradient boosting tree (XGBoost), Adaptive Boosting (Adaboost), and artificial neural network (ANN) based on 21 clinical biochemical indicators and features. Evaluation metrics included accuracy (ACC), precision (PRE), specificity (SPE), sensitivity (SEN) or regression rate(Recall), and f1 score. Evaluation metrics for the models primarily included ACC, PRE, SPE, SEN or Recall, f1 score, and area under the curve(AUC). Evaluation metrics were visualized using confusion matrices and ROC curves. Among the ensemble learning methods, RF, XGBoost, and Adaboost performed the best. RF achieved a training dataset score of 0.769 (95% CI: 0.759-0.835) and a testing dataset score of 0.755 (95% CI: 0.660-0.760) (AUC: 0.786, 95%CI: 0.722-0.803), while XGBoost achieved a training dataset score of 0.755 (95% CI: 95%CI: 0.711-0.809) and a testing dataset score of 0.745 (95% CI: 0.660-0.764) (AUC: 0.777, 95% CI: 0.726-0.798). Adaboost scored 0.789 on the training dataset (95% CI: 0.782-0.857) and 0.774 on the testing dataset (95% CI: 0.651-0.774) (AUC: 0.799, 95% CI: 0.703-0.802). In terms of feature importance (FI) in ensemble learning, Bone metastases at first visit, prostatic volume, age, and T1-T2 have significant proportions in RF's FI. fPSA, TPSA, and tumor burden have significant proportions in Adaboost's FI, while f/TPSA, LDH, and testosterone have the highest proportions in XGBoost. Our findings indicate that ensemble learning methods demonstrate good performance in classifying HSPC patient data, with TNM staging and fPSA being important classification indicators. These discoveries provide valuable references for distinguishing different Gleason scores, facilitating more accurate patient assessments and personalized treatment plans.

Prostate hyperplasia in St Mary's Hospital Lacor: utility of prostate specific antigen in screening for prostate malignancy

INTRODUCTION

Prostate cancer is the second commonest cancer in men worldwide. At present, every patient with lower urinary tract symptoms (LUTS) in St. Mary's Hospital Lacor is undergoing prostate biopsy regardless of the prostate specific antigen (PSA) level. We sought to determine the association between PSA and malignant prostate histology.

METHODS

This was a retrospective study. Data on age, PSA, prostate volume and prostate histology reported between Jan 2012 and Dec 2019 were retrieved from St. Mary's Hospital Lacor archive and analyzed using STATA SE/13.0.

RESULTS

Records of 97 patients with LUTS was analyzed. The median (range) age of the patients was 71 (43-100) years. Median (range) of prostate volume was 91.8 (8.0-360.0) cc. Overall, PSA ranged from 0.21 to 399.2 ng/ml. Prostate histology showed 3.1% acinar adenocarcinoma, 24.7% adenocarcinoma and 72.2% benign prostatic hyperplasia. The median PSA amongst patients with malignant and non-malignant prostates were 15.8 ng/ml and 6.07 ng/ml respectively. Serum PSA level was significantly higher in patients with malignant prostate histology (Difference of mean= 9.7; p=0.001).

CONCLUSION

Patients with LUTS and PSA levels of 15ng/ml or more were more likely to have malignant prostate histology.

Suicide risk among prostate cancer patients before and after the implementation of prostate-specific antigen-based prostate screening in Lithuania in 2006

Despite good prognosis, increased suicide rates are reported for prostate cancer. The aim of this study was to assess the risk of suicide among prostate cancer patients before and after the start of nation-wide prostate-specific antigen (PSA)-based screening programme. Prostate cancer cases diagnosed between 2000 and 2011 were identified from the population-based Lithuanian Cancer Registry and analysis was conducted in 2018. Analysis was stratified by period of diagnosis, age, Gleason score, extent of disease, and time since diagnosis. Standardized mortality ratios (SMRs) were used to assess suicide risk. SMRs were calculated by dividing the observed number of suicides among prostate cancer patients by the expected number of suicides from the general population. Overall, 25 786 prostate cancer cases were diagnosed 2000-2011, and 135 suicides occurred among them compared with expected number of 133 (SMR: 1.10; 95% confidence interval (CI) 0.85-1.20). The suicide risk among prostate cancer patients was 1.08 before and 0.97 after the start of nation-wide PSA-based screening programme. Statistically significant increase in suicide risk was associated with Gleason score 8-10 in the prescreening period (SMR: 2.45; 95% CI 1.23-4.90). Suicide risk among prostate cancer patients before and after introduction of nation-wide PSA-based screening programme is similar to that in the general population.

Gleason score 7 prostate cancer on needle biopsy: relation of primary pattern 3 or 4 to pathological stage and progression after radical prostatectomy

PURPOSE

There have been only a few contradictory publications assessing whether Gleason score 4 + 3 = 7 has a worse prognosis than 3 + 4 = 7 on biopsy material in predicting pathological stage and biochemical recurrence. Older studies predated the use of the modified Gleason grading system established in 2005.

MATERIALS AND METHODS

We retrospectively studied 1,791 cases of Gleason score 7 on prostatic biopsy to determine whether the breakdown of Gleason score 7 into 3 + 4 vs 4 + 3 has prognostic significance in the modern era.

RESULTS

There was no difference in patient age, preoperative serum prostate specific antigen, maximum tumor percent per core or the number of positive cores between Gleason score 3 + 4 = 7 and Gleason score 4 + 3 = 7. Gleason score 4 + 3 = 7 showed an overall correlation with pathological stage (organ confined, focal extraprostatic extension, nonfocal extraprostatic extension, seminal vesicle invasion/lymph node metastases, p = 0.005). On multivariate analysis Gleason score 4 + 3 = 7 (p = 0.03), number of positive cores (p = 0.002), maximum percent of cancer per core (p = 0.006) and preoperative serum prostate specific antigen (p = 0.03) all correlated with pathological stage. Gleason score 4 + 3 = 7 on biopsy was also associated with an increased risk of biochemical progression after radical prostatectomy (p = 0.0001). On multivariate analysis Gleason score 4 + 3 = 7 (p = 0.001), maximum percent of cancer per core (p <0.0001) and preoperative serum prostate specific antigen (p <0.0001) but not number of positive cores correlated with the risk of biochemical progression after radical prostatectomy.

CONCLUSIONS

Our study further demonstrates that Gleason score 7 should not be considered a homogenous group for the purposes of disease management and prognosis.

Current perspectives on Gleason grading of prostate cancer

The 5-tier Gleason grading system for prostate cancer, introduced in 1966, has been proven to be one of the main independent predictors of prostate cancer outcome. This review addresses interobserver concordance in Gleason grading; the persistence of grading discrepancies with frequent upgrading from the biopsy to the prostatectomy specimen; the 2005 International Society of Urologic Pathologists' modifications to Gleason grading; the impact of this modified grading on grade migration and outcome prediction; and molecular correlates of cancer morphology. Data from the most recent years are emphasized.

Undergrading and understaging in patients with clinically insignificant prostate cancer who underwent radical prostatectomy

PURPOSE

The aim of our study is to evaluate the undergrading and understaging rates in patients with clinically localized insignificant prostate cancer who underwent radical prostatectomy.

MATERIALS AND METHODS

Between July 2005 and July 2008, 406 patients underwent radical prostatectomy for clinical localized prostate cancer in our hospital. Based on preoperative data, 93 of these patients fulfilled our criteria of non-significance: Gleason score < 7, stage T1c, PSA < 10 ng/mL and percentage of affected fragments less than 25%. The pathologic stage and Gleason score were compared to preoperative data to evaluate the rate of understaging and undergrading. The biochemical recurrence free survival of these operated insignificant cancers were also evaluated.

RESULTS

On surgical specimen analysis 74.7% of patients had Gleason score of 6 or less and 25.3% had Gleason 7 or greater. Furthermore 8.3% of cases showed extracapsular extension. After 36 months of follow-up 3.4% had biochemical recurrence, defined by a PSA above 0.4 ng/mL.

CONCLUSIONS

Despite the limited number of cases, we have found considerable rates of undergrading and understaging in patients with prostate cancer whose current definitions classified them as candidates for active surveillance. According to our results the current definition seems inadequate as up to a third of patients had higher grade or cancer outside the prostate.

TMPRSS2-ERG gene fusion is associated with low Gleason scores and not with high-grade morphological features

TMPRSS2-ERG gene rearrangement is seen in about half of clinically localized prostate cancers, yet controversy exists with regard to its prognostic implications. Similarly, the relationship of TMPRSS2-ERG fusion to Gleason score and morphology remains uncertain. We assigned Gleason scores and recorded morphological features for 521 clinically localized prostate cancers sampled in triplicate and arrayed in eight tissue microarray blocks. Fluorescence in situ hybridization was performed to delineate TMPRSS2-ERG aberrations. Using maximum Gleason score, based on three core evaluation, and overall Gleason score, based on prostatectomy sections, Fisher's exact test was performed for tumors with TMPRSS2-ERG translocation/deletion, copy number increase (≥ 3) of the TMPRSS2-ERG region without translocation/deletion, and copy number increase and concomitant translocation/deletion. In all, 217 (42%) translocation/deletion and 30 (5.9%) copy number increase-alone cases were detected. Among 217 translocation/deletion cases, 32 had translocation/deletion with copy number increase. In all, 237, 200, and 75 cancers had maximum core-specific Gleason score of 6, 7, and 8-10, respectively. Tumors with translocation/deletion tended toward lower Gleason scores than those without (P=0.002) with similar results for overall Gleason score (P=0.02); copy number increase cases tended toward higher Gleason scores than those without (P<0.001). Gleason score of 8-10 tumors demonstrated lower odds of translocation/deletion (odds ratio (OR) 0.38; 95% CI 0.21-0.68) and higher odds of copy number increase alone (OR 7.33; 95% CI 2.65-20.31) or copy number increase+translocation/deletion (OR 3.03; 95% CI 1.12-8.15) relative to Gleason score of <7 tumors. No significant difference in TMPRSS2-ERG incidence was observed between patients with and without cribriform glands, glomerulations, signet-ring cells, or intraductal cancer (P=0.821, 0.095, 0.132, 0.375). TMPRSS2-ERG gene fusion is associated with lower core-specific and overall Gleason scores and not with high-grade morphologies. Conversely, TMPRSS2-ERG copy number increase, with or without rearrangement, is associated with higher Gleason score. These findings indicate that translocation/deletion of TMPRSS2-ERG is not associated with histological features of aggressive prostate cancer.

Determining relative importance of variables in developing and validating predictive models

Background

Multiple regression models are used in a wide range of scientific disciplines and automated model selection procedures are frequently used to identify independent predictors. However, determination of relative importance of potential predictors and validating the fitted models for their stability, predictive accuracy and generalizability are often overlooked or not done thoroughly.

Methods

Using a case study aimed at predicting children with acute lymphoblastic leukemia (ALL) who are at low risk of Tumor Lysis Syndrome (TLS), we propose and compare two strategies, bootstrapping and random split of data, for ordering potential predictors according to their relative importance with respect to model stability and generalizability. We also propose an approach based on relative increase in percentage of explained variation and area under the Receiver Operating Characteristic (ROC) curve for developing models where variables from our ordered list enter the model according to their importance. An additional data set aimed at identifying predictors of prostate cancer penetration is also used for illustrative purposes.

Results

Age is chosen to be the most important predictor of TLS. It is selected 100% of the time using the bootstrapping approach. Using the random split method, it is selected 99% of the time in the training data and is significant (at 5% level) 98% of the time in the validation data set. This indicates that age is a stable predictor of TLS with good generalizability. The second most important variable is white blood cell count (WBC). Our methods also identified an important predictor of TLS that was otherwise omitted if relying on any of the automated model selection procedures alone. A group at low risk of TLS consists of children younger than 10 years of age, without T-cell immunophenotype, whose baseline WBC is < 20 × 10⁹/L and palpable spleen is < 2 cm. For the prostate cancer data set, the Gleason score and digital rectal exam are identified to be the most important indicators of whether tumor has penetrated the prostate capsule.

Conclusion

Our model selection procedures based on bootstrap re-sampling and repeated random split techniques can be used to assess the strength of evidence that a variable is truly an independent and reproducible predictor. Our methods, therefore, can be used for developing stable and reproducible models with good performances. Moreover, our methods can serve as a good tool for validating a predictive model. Previous biological and clinical studies support the findings based on our selection and validation strategies. However, extensive simulations may be required to assess the performance of our methods under different scenarios as well as check their sensitivity to a random fluctuation in the data.