Development, Validation And Optimization Of Individualized Clinical Prediction Models In Genitourinary Cancers

Part1. Development of Clinical Prediction Model in The Form of NomogramNomogram based on Logistic regression modelPURPOSE:Optimal management for penile cancer in patients with clinically negative lymph nodes is still under debate. We developed and evaluated a nomogram to stratify patients who are suitable candidates for further treatment.MATERIALS AND METHODS:This study included110men with penile cancer and clinically negative lymph nodes from1990to2008. All patients underwent primary tumor resection and regional lymphadenectomy. We retrospectively reviewed medical records and tumor slides. Statistical analysis was done in R with library rms.RESULTS:The lymph node metastasis rate in the entire cohort was23.6%. The final model, presented as a nomogram, included T stage, grade, lymphovascular invasion and p53expression. Only lymphovascular invasion showed independent prognostic value on multivariate analysis (p=0.024). The model also showed good calibration (bootstrap corrected concordance index0.79). To determine the clinical usefulness of the nomogram we compared it with the European Association of Urology risk classification using decision curve analysis. At a10%probability threshold our nomogram led to1positive result per100patients without an increase in the number of false-positive results. At this probability threshold the model also decreased13unnecessary interventions per100patients without missing more metastatic disease.CONCLUSIONS:We generated a nomogram in patients with clinically node negative penile cancer based on readily available pathological factors. The clinical usefulness of the nomogram was evidenced by decision curve analysis. Nomogram based on Cox proportional hazards modelPURPOSE:We developed a nomogram to predict the duration of drainage in patients with penile cancer treated with inguinal lymph node dissection. MATERIALS AND METHODS:A total of111groin basins in56patients who underwentradical inguinal lymph node dissection for penile cancer were retrospectively assessed. We retrieved the clinicopathological factors from the medical records including age, body mass index, albumin, smoking history, hypertension, diabetes, preoperative radiotherapy/chemotherapy, palpable lymph nodes, previous lymph node biopsy, total number of resected lymph nodes and ratio of positive lymph nodes. The criterion of drain removal was total drain output of50ml or less per day for2days starting from postoperative day3. A multivariate Cox proportional hazards model was used to explore the risk factors of drainage duration and variable selection was performed according to Akaike’s information criteria. A nomogram was built based on regression coefficients and internally validated with200bootstrap resamples.RESULTS:Median postoperative drainage duration was7days. The prediction model using pretreatment factors showed a concordance index of0.55. With the addition of lymph node related variables a second model was constructed which produced a better concordance index (0.65) and good calibration. On multivariate analysis young age, high body mass index, total number of resected lymph nodes and ratio of positive lymph nodes were independent predictors of prolonged lymphatic drainage.CONCLUSIONS:On the basis of readily obtained clinicopathological variables we developed a nomogram to predict the duration of lymphatic drainage which, if externally validated, could be helpful for patient consultation, treatment decision making and clinical trial design. Part2. Comprehensive Evaluation of Clinical Prediction ModelComprehensive Evaluation of Clinical Prediction Model Based on Logistic Regression PURPOSE:Several prediction models have been developed to estimate the outcomes of prostate biopsies. Most of these tools were designed for use with Western populations and have not been validated across different ethnic groups. Therefore, we evaluated the predictive value of the Prostate Cancer Prevention Trial (PCPT) and the European Randomized Study of Screening for Prostate Cancer (ERSPC) risk calculators in a Chinese cohort.MATERIALS AND METHODS:Clinicopathological information was obtained from495Chinese men who had undergone extended prostate biopsies between January2009and March2011. The estimated probabilities of prostate cancer and high-grade disease (Gleason>6) were calculated using the PCPT and ERSPC risk calculators. Overall measures, discrimination, calibration and clinical usefulness were assessed for the model evaluation. RESULTS:Of these patients,28.7%were diagnosed with prostate cancer and19.4%had high-grade disease. Compared to the PCPT model and the prostate-specific antigen (PSA) threshold of4ng ml(-1), the ERSPC risk calculator exhibited better discriminative ability for predicting positive biopsies and high-grade disease (the area under the curve was0.831and0.852, respectively, P<0.01for both). Decision curve analysis also suggested the favourable clinical utility of the ERSPC calculator in the validation dataset. Both prediction models demonstrated miscalibration:the risk of prostate cancer and high-grade disease was overestimated by approximately20%for a wide range of predicted probabilities. CONCLUSIONS:The ERSPC risk calculator outperformed both the PCPT model and the PSA threshold of4ng ml(-1) in predicting prostate cancer and high-grade disease in Chinese patients. However, the prediction tools derived from Western men significantly overestimated the probability of prostate cancer and high-grade disease compared to the outcomes of biopsies in a Chinese cohort. Comprehensive Evaluation of Clinical Prediction Model Based on Cox ModelPURPOSE:Using population-based cancer registry, Thuret and colleagues had developed3nomograms for estimating cancer-specific mortality (CSM) in men with penile squamous cell carcinoma. In initial cohort, only23.0%of the patients were treated with inguinal lymphadenectomy and had pN stage. To generalize the prediction models in clinical practice, we evaluated the performance of the3nomograms in a series of penile cancer patients treated with definitive surgery.MATERIALS AND METHODS:Clinicopathological information was obtained from160MO penile cancer patients who underwent primary tumor excision and regional lymphadenectomy between1990and2008. The predicted probabilities of CSM were calculated from3nomograms which were based on different disease stage definitions and tumor grade (TG). Discrimination, calibration, and clinical usefulness were assessed for comparisons of model performance.RESULTS:The discrimination ability was similar in nomograms using TNM classification or American Joint Committee on Cancer stage (Harrell’s concordance index=0.817and0.832, respectively), while was inferior in prediction model included Surveillance, Epidemiology and End Results stage (Harrell’s concordance index=0.728). A better agreement with observed CSM was shown for the model consisted of TNM classification and TG, which also achieved favorable clinical net benefit with threshold probability in a range from O to42%.CONCLUSIONS:The nomogram consisted of TNM classification and TG was shown to have better performance for predicting CSM in penile cancer patient received definitive surgery. Our data support the integration of this model in decision making and trial design. Part3. Assessment of the Incremental Value of New Prognostic FactorAssessment of the Impact of New Prognostic factor on Binary OutcomePURPOSE:Accurate assessment of disease characteristics is a prerequisite for treatment decision making regarding small renal masses. In this study we evaluate the association between visceral obesity and Fuhrman grade in patients with cTla renal cell carcinoma.MATERIALS AND METHODS:We retrospectively collected data on186patients with surgically treated cT1a renal cell carcinoma. Single slice computerized tomography was used to measure the area of visceral and subcutaneous adipose tissue. Visceral obesity was calculated as the proportion of visceral adipose tissue to overall adipose tissue. Other analyzed factors included clinical characteristics (age, gender, body mass index and tumor size) and anatomical features of the tumor defined by the R.E.N.A.L nephrometry score. The association between predictors and high grade disease (Fuhrman grade Ⅲ or Ⅳ) were assessed using logistic regression analyses.RESULTS:A total of47(25.3%) tumors were classified as high grade. The percentage of visceral adipose tissue was higher in male participants but did not correlate with body mass index, age or tumor size. In univariate analyses the percentage of visceral adipose tissue and tumor size were significantly associated with higher Fuhrman grade. Multivariate analysis showed that the percentage of visceral adipose tissue (OR1.06,.p=0.0018) and tumor size (OR1.91, p=0.047) were independent predictors of high grade cancer. Addition of the percentage of visceral adipose tissue to a model including clinical characteristics and anatomical features of the tumor remarkably improved its discriminatory ability (p=0.0010).CONCLUSIONS:Increased visceral obesity was found to be strongly associated with higher Fuhrman grade in patients with cT1a renal cell carcinoma. Further studies are needed to confirm these findings and discover the underlying biological mechanism. Assessment of the Impact of New Prognostic factor on Survival Outcome PURPOSE:We determined whether the new N staging system, which was introduced in2009, leads to more specific prediction of survival in patients with penile squamous cell carcinoma.MATERIALS AND METHODS:We analyzed the records of60patients in whom node positive penile cancer was surgically resected from1990to2008. All cases were staged according the6th and7th N staging system after pathological review. Histopathological information on the number of positive lymph nodes, lymph node metastasis laterality, extranodal extension, pelvic lymph node metastasis and lymph node ratio were also recorded. We evaluated the added information on these nodal related prognostic factors to the current N classification. Recurrence-free survival was calculated. Predictive accuracy was assessed by the concordance index.RESULTS:Disease recurred in27of the60patients (42.4%) at a median of10months. In the33patients without recurrence at the last visit median followup was53months. Using the6th N classification the3-year recurrence-free survival rate was69.8%,48.2%and33.3%for the N1, N2and N3categories, respectively. Log rank survival analysis failed to show a statistical difference (p=0.054). For the new7th N categories the3-year recurrence-free survival rate was87.5%,57%and31.8%in the corresponding N1to N3groups. Better survival stratification was observed on analysis (p<0.001). Adding lymph node metastasis laterality or lymph node ratio significantly increased the accuracy of the7th N category to predict recurrence-free survival.CONCLUSIONS:The new N staging system better reflects the prognosis in patients with penile cancer.