A nomogram is more accurate than a regression tree in predicting lymph node invasion in prostate cancer

The predictive value of machine learning and nomograms for lymph node metastasis of prostate cancer: a systematic review and meta-analysis

BACKGROUND

In clinical practice, there are currently a variety of nomograms for predicting lymph node metastasis (LNM) of prostate cancer. At the same time, some scholars have introduced machine learning (ML) into the prediction of LNM of prostate cancer. However, the predictive value of nomograms and ML remains controversial. Based on this situation, this systematic review and meta-analysis was performed to explore the predictive value of various nomograms currently recommended and newly-developed ML models for LNM in prostate cancer patients.

EVIDENCE ACQUISITION

Cochrane, PubMed, Embase, and Web of Science were searched up to November 1, 2022. The risk of bias in the included studies was evaluated using the Prediction model Risk of Bias Assessment Tool (PROBAST). The concordance index (C-index), sensitivity, and specificity were adopted to evaluate the predictive accuracy of the models.

RESULTS

Thirty-one studies (18,803 patients) were included. Seven kinds of nomograms currently recommended, dominated by Briganti nomogram or MSKCC nomogram, were covered in the included studies. For newly-developed ML models, the C-index for LNM prediction in the training set and validation set was 0.846 [95%CI (0.818, 0.873)] and 0.862 [95%CI (0.819-0.905)] respectively. Most ML models in the training set were based on Logistic Regression (LR), which had a sensitivity of 0.78 [95%CI (0.70, 0.85)] and a specificity of 0.85 [95%CI (0.77, 0.90)] in the training set, and a sensitivity of 0.81 [95%CI (0.67, 0.89)] and a specificity of 0.82 [95%CI (0.75, 0.88)] in the validation set. For the recommended nomograms, the C-index in the validation set was 0.745 [95%CI (0.701, 0.790)] for the Briganti nomogram and 0.714 [95%CI (0.662, 0.765)] for the MSKCC nomogram.

CONCLUSION

The predictive accuracy of ML is superior to existing clinically recommended nomograms, and appropriate updates can be conducted to existing nomograms according to special situations.

A machine learning-assisted decision-support model to better identify patients with prostate cancer requiring an extended pelvic lymph node dissection

OBJECTIVES

To develop a machine learning (ML)-assisted model to identify candidates for extended pelvic lymph node dissection (ePLND) in prostate cancer by integrating clinical, biopsy, and precisely defined magnetic resonance imaging (MRI) findings.

PATIENTS AND METHODS

In all, 248 patients treated with radical prostatectomy and ePLND or PLND were included. ML-assisted models were developed from 18 integrated features using logistic regression (LR), support vector machine (SVM), and random forests (RFs). The models were compared to the Memorial SloanKettering Cancer Center (MSKCC) nomogram using receiver operating characteristic-derived area under the curve (AUC) calibration plots and decision curve analysis (DCA).

RESULTS

A total of 59/248 (23.8%) lymph node invasions (LNIs) were identified at surgery. The predictive accuracy of the ML-based models, with (+) or without (-) MRI-reported LNI, yielded similar AUCs (RFs⁺ /RFs^- : 0.906/0.885; SVM⁺ /SVM^- : 0.891/0.868; LR⁺ /LR^- : 0.886/0.882) and were higher than the MSKCC nomogram (0.816; P < 0.001). The calibration of the MSKCC nomogram tended to underestimate LNI risk across the entire range of predicted probabilities compared to the ML-assisted models. The DCA showed that the ML-assisted models significantly improved risk prediction at a risk threshold of ≤80% compared to the MSKCC nomogram. If ePLNDs missed was controlled at <3%, both RFs⁺ and RFs^- resulted in a higher positive predictive value (51.4%/49.6% vs 40.3%), similar negative predictive value (97.2%/97.8% vs 97.2%), and higher number of ePLNDs spared (56.9%/54.4% vs 43.9%) compared to the MSKCC nomogram.

CONCLUSIONS

Our ML-based model, with a 5-15% cutoff, is superior to the MSKCC nomogram, sparing ≥50% of ePLNDs with a risk of missing <3% of LNIs.

MRI and 11C Acetate PET/CT for Prediction of Regional Lymph Node Metastasis in Newly Diagnosed Prostate Cancer

Background

The aim of the study was to examine the value of quantitative and qualitative MRI and ¹¹C acetate PET/CT parameters in predicting regional lymph node (LN) metastasis of newly diagnosed prostate cancer (PCa).

Patients and methods

Patients with intermediate (n = 6) and high risk (n = 47) PCa underwent 3T MRI (40 patients) and ¹¹C acetate PET/CT (53 patients) before extended pelvic LN dissection. For each patient the visually most suspicious LN was assessed for mean apparent diffusion coefficient (ADCmean), maximal standardized uptake value (SUVmax), size and shape and the primary tumour for T stage on MRI and ADCmean and SUVmax in the index lesion. The variables were analysed in simple and multiple logistic regression analysis.

Results

All variables, except ADCmean and SUVmax of the primary tumor, were independent predictors of LN metastasis. In multiple logistic regression analysis the best model was ADCmean in combintion with MRI T-stage where both were independent predictors of LN metastasis, this combination had an AUC of 0.81 which was higher than the AUC of 0.65 for LN ADCmean alone and the AUC of 0.69 for MRI T-stage alone.

Conclusions

Several quantitative and qualitative imaging parameters are predictive of regional LN metastasis in PCa. The combination of ADCmean in lymph nodes and T-stage on MRI was the best model in multiple logistic regression with increased predictive value compared to lymph node ADCmean and T-stage on MRI alone.

Global methylation profiling for risk prediction of prostate cancer

PURPOSE

The aim of this study was to investigate the promoter hypermethylation as diagnostic markers to detect malignant prostate cells and as prognostic markers to predict the clinical recurrence of prostate cancer.

EXPERIMENTAL DESIGN

DNA was isolated from prostate cancer and normal adjacent tissues. After bisulfite conversion, methylation of 14,495 genes was evaluated using the Methylation27 microarrays in 238 prostate tissues. We analyzed methylation profiles in four different groups: (i) tumor (n = 198) versus matched normal tissues (n = 40), (ii) recurrence (n = 123) versus nonrecurrence (n = 75), (iii) clinical recurrence (n = 80) versus biochemical recurrence (n = 43), and (iv) systemic recurrence (n = 36) versus local recurrence (n = 44). Group 1, 2, 3, and 4 genes signifying biomarkers for diagnosis, prediction of recurrence, clinical recurrence, and systemic progression were determined. Univariate and multivariate analyses were conducted to predict risk of recurrence. We validated the methylation of genes in 20 independent tissues representing each group by pyrosequencing.

RESULTS

Microarray analysis revealed significant methylation of genes in four different groups of prostate cancer tissues. The sensitivity and specificity of methylation for 25 genes from 1, 2, and 4 groups and 7 from group 3 were shown. Validation of genes by pyrosequencing from group 1 (GSTP1, HIF3A, HAAO, and RARβ), group 2 (CRIP1, FLNC, RASGRF2, RUNX3, and HS3ST2), group 3 (PHLDA3, RASGRF2, and TNFRSF10D), and group 4 (BCL11B, POU3F3, and RASGRF2) confirmed the microarray results.

CONCLUSIONS

Our study provides a global assessment of DNA methylation in prostate cancer and identifies the significance of genes as diagnostic and progression biomarkers of prostate cancer.

Formalized prediction of clinically significant prostate cancer: is it possible?

Greater understanding of the biology and epidemiology of prostate cancer in the last several decades have led to significant advances in its management. Prostate cancer is now detected in greater numbers at lower stages of disease and is amenable to multiple forms of efficacious treatment. However, there is a lack of conclusive data demonstrating a definitive mortality benefit from this earlier diagnosis and treatment of prostate cancer. It is likely due to the treatment of a large proportion of indolent cancers that would have had little adverse impact on health or lifespan if left alone. Due to this overtreatment phenomenon, active surveillance with delayed intervention is gaining traction as a viable management approach in contemporary practice. The ability to distinguish clinically insignificant cancers from those with a high risk of progression and/or lethality is critical to the appropriate selection of patients for surveillance protocols versus immediate intervention. This chapter will review the ability of various prediction models, including risk groupings and nomograms, to predict indolent disease and determine their role in the contemporary management of clinically localized prostate cancer.

Mapping of lymphatic drainage from the prostate using filtered 99mTc-sulfur nanocolloid and SPECT/CT

We have developed a practice procedure for prostate lymphoscintigraphy using SPECT/CT and filtered (99m)Tc-sulfur nanocolloid, as an alternative to the proprietary product (99m)Tc-Nanocoll, which is not approved in the United States.

METHODS

Ten patients were enrolled for this study, and all received radiotracer prepared using a 100-nm membrane filter at a commercial radiopharmacy. Whole-body scans and SPECT/CT studies were performed within 1.5-3 h after the radiotracer had been administered directly into 6 locations of the prostate gland under transrectal ultrasound guidance. The radiation dose was estimated from the first 3 patients. Lymphatic drainage mapping was performed, and lymph nodes were identified.

RESULTS

The estimated radiation dose ranged from 3.9 to 5.2 mSv/MBq. The locations of lymph nodes draining the prostate gland were similar to those found using the proprietary product.

CONCLUSION

When the proprietary radiolabeled nanocolloid indicated for lymphoscintigraphy is not available, prostate lymph node mapping and identification are still feasible using filtered (99m)Tc-sulfur nanocolloid.

Clinical importance of lymph node density in predicting outcome of prostate cancer patients

BACKGROUND

To evaluate the prognostic role of lymph node density (LND) in patients affected by prostate cancer (PCa) and treated with radical prostatectomy (RP), pelvic lymph node dissection (PLND), and adjuvant hormonal blockade.

METHODS

A total of 124 consecutive patients with lymph node positive PCa formed the basis of this report. Clinical and pathological parameters were collected. All patients were stratified in two groups according to LND: Group A (LND ≤ 32%) and Group B (LND ≥ 33%). Time to the first biochemical recurrence (BCR) was the main measure of outcome.

RESULTS

Mean number of lymph nodes removed was 5.2 (range 1-17). The mean number of positive lymph nodes was 1.6 (range 1-5). At a mean follow-up of 84.3 mo, 22 patients in Group A (43.1%) and 40 in Group B (54.7%) had BCR. The mean overall BCR-free survival was 46.2 mo (range 8-90). No significant correlation was found between the number of positive lymph nodes and BCR-free survival (P = 0.68). In addition, the patients with LND ≥ 33% had a poor prognosis with significantly decreased disease-specific and BCR-free survival rates (hazard ratio: 0.48; 95% CI, 0.28-0.78; P = 0.0039). At multivariate and artificial neural network (ANN) analyses, LND, Gleason score, and stage were identified as independent prognostic factors of BCR-free survival (P = 0.002; P = 0.003; P = 0.003).

CONCLUSIONS

The current study highlights the role of LND in predicting BCR-free survival in patients with lymph node positive PCa after prostatectomy. It also reinforces the need for a stratification of patients with nodal metastasis.

[Highlights of the 2010 ERUS meeting, September 29 to October 1, 2010, Bordeaux, France]

This work summarizes the highlights of what was presented at the seventh edition of the European Robotic Urology Symposium meeting which took place in Bordeaux, France, from September 29 to October 1, 2010. Future developments of robotic surgery and training in robotic were discussed. Robotic assisted laparoscopic radical prostatectomy was largely discussed. The use of robotic in renal and bladder surgery was also developed. The congress contained update lectures, debates, live cases transmission of robotic surgery, and poster and video communications.

Should all men having a radical prostatectomy have a pelvic lymph node dissection? No

Collapse

Predictive and prognostic models in radical prostatectomy candidates: a critical analysis of the literature

CONTEXT

Numerous predictive and prognostic tools have recently been developed for risk stratification of prostate cancer (PCa) patients who are candidates for or have been treated with radical prostatectomy (RP).

OBJECTIVE

To critically review the currently available predictive and prognostic tools for RP patients and to describe the criteria that should be applied in selecting the most accurate and appropriate tool for a given clinical scenario.

EVIDENCE ACQUISITION

A review of the literature was performed using the Medline, Scopus, and Web of Science databases. Relevant reports published between 1996 and January 2010 identified using the keywords prostate cancer, radical prostatectomy, predictive tools, predictive models, and nomograms were critically reviewed and summarised.

EVIDENCE SYNTHESIS

We identified 16 predictive and 22 prognostic validated tools that address a variety of end points related to RP. The majority of tools are prediction models, while a few consist of risk-stratification schemes. Regardless of their format, the tools can be distinguished as preoperative or postoperative. Preoperative tools focus on either predicting pathologic tumour characteristics or assessing the probability of biochemical recurrence (BCR) after RP. Postoperative tools focus on cancer control outcomes (BCR, metastatic progression, PCa-specific mortality [PCSM], overall mortality). Finally, a novel category of tools focuses on functional outcomes. Prediction tools have shown better performance in outcome prediction than the opinions of expert clinicians. The use of these tools in clinical decision-making provides more accurate and highly reproducible estimates of the outcome of interest. Efforts are still needed to improve the available tools' accuracy and to provide more evidence to further justify their routine use in clinical practice. In addition, prediction tools should be externally validated in independent cohorts before they are applied to different patient populations.

CONCLUSIONS

Predictive and prognostic tools represent valuable aids that are meant to consistently and accurately provide most evidence-based estimates of the end points of interest. More accurate, flexible, and easily accessible tools are needed to simplify the practical task of prediction.

A new formula for prostate cancer lymph node risk

INTRODUCTION

The successful treatment of prostate cancer depends on the accurate estimation of the risk of regional lymph node (LN) involvement. The Roach formula (RF) has been criticized as overestimating LN risk. A modification of the RF has been attempted by other investigators using simplified adjustment ratios: the Nguyen formula (NF).

METHODS AND MATERIALS

The National Cancer Institute Surveillance, Epidemiology, and End Results database was investigated for patients treated in 2004 through 2006 for whom at least 10 LN were examined at radical prostatectomy, cT1c or cT2 disease, and prostate-specific antigen (PSA) <26 ng/ml (N = 2,930). The Yale formula (YF) was derived from half of the sample (n = 1,460), and validated in the other half (n = 1,470).

RESULTS

We identified 2,930 patients. Only 4.6% of patients had LN+, and 72.6% had cT1c disease. Gleason (GS) 8-10 histology was found in 14.4% of patients. The YF for prediction of %LN+ risk is [GS - 5] × [PSA/3 + 1.5 × T], where T = 0, 1, and 2 for cT1c, cT2a, and cT2b/cT2c. Within each strata of predicted %LN+ risk, the actual %LN+ was closest to the YF. Using a >15% risk as an indicator of high-risk disease, the YF had increased sensitivity (39.0% vs. 13.6%) compared with the NF, without a significant reduction in specificity (94.9% vs. 98.8%). The NF was overly restrictive of the high-risk group, with only 2% of patients having a >15% risk of LN+ by that formula.

CONCLUSION

The YF performed better than the RF and NF and was best at differentiating patients at high risk for LN+ disease.

A nomogram for predicting overall survival of women with endometrial cancer following primary therapy: toward improving individualized cancer care

OBJECTIVES

Traditionally we have relied mainly on final FIGO stage to estimate overall oncologic outcome in endometrial cancer patients. However, it is well known that other patient factors may play equally important roles in outcome. Our objective was to develop a clinically useful nomogram in the hope of providing a more individualized and accurate estimation of overall survival (OS) following primary therapy.

METHODS

Using a prospectively maintained endometrial cancer database, 1735 patients treated between 1993 and 2008 were analyzed. Characteristics known to predict OS were collected. For each patient, points were assigned to each of these 5 variables. A total score was calculated. The association between each predictor and the outcome was assessed by multivariable modeling. The corresponding 3-year OS probabilities were then determined from the nomogram.

RESULTS

The median age was 62 years (range, 25-96). Final grade included: G1 (471), G2 (622), G3 (634), and missing (8). Stage included: IA (501), IB (590), IC (141), IIA (36), IIB (75), IIIA (116), IIIB (6), IIIC (135), IVA (7), and IVB (128). Histology included: adenocarcinoma (1376), carcinosarcoma (100), clear cell (62), and serous (197). Median follow-up for survivors was 29.2 months (0-162.2 months). Concordance probability estimator for the nomogram is 0.746+/-0.011.

CONCLUSION

We developed a nomogram based on 5 easily available clinical characteristics to predict OS with a high concordance probability. This nomogram incorporates other individualized patient variables beyond FIGO stage to more accurately predict outcome. This new tool may be useful to clinicians in assessing patient risk when deciding on follow-up strategies.

Novel predictive tools for Irish radical prostatectomy pathological outcomes: development and validation

AIMS

We developed and validated prostate cancer predictive models for Irish patients, allowing individualised predictions of radical prostatectomy pathological outcomes.

METHODS

Retrospective review of the Irish Prostate Cancer Research Consortium database from 2003 to 2008 was performed. Two predictive models were formulated: a replica of the Partin tables (n = 169) and a look-up table based on PSA and biopsy Gleason Score (n = 253). Clinico-pathological parameters were compared to the Partin data set. Internal validation was performed.

RESULTS

In total, 70% of patients were at clinical stage T1c. 5.8% had a PSA less than 4.1 ng/ml, whereas 25% of the Partin patients had a PSA in this range. Maximal predictive accuracy was seen for seminal vesicle invasion (area under the curve = 72%). Prediction of extra-prostatic extension and lymph node involvement was only equivalent to that of a chance phenomenon.

CONCLUSIONS

Our current results do not support the introduction of the formulated predictive models into routine clinical practice.

Prospective comparison of five mediators of the systemic response after high-intensity focused ultrasound and targeted cryoablation for localized prostate cancer

OBJECTIVE

To evaluate whether high-intensity focused ultrasound (HIFU) is less invasive than targeted cryoablation of the prostate (TCAP), as experimental studies suggest that the acute-phase reaction is proportional to surgery-induced tissue damage.

PATIENTS AND METHODS

Between May 2004 and December 2007, 127 consecutive patients undergoing HIFU (71) or TCAP (56) in our departments were assessed prospectively. Blood samples were collected 24 h before (T0), during surgery (T1), at the end of anaesthesia (T2), and at 12 (T3), 24 (T4) and 36 h after surgery (T5). The extent of the systemic response to surgery-induced tissue trauma was measured by assessing the levels of acute-phase markers tumour necrosis factor-alpha (TNF-alpha), interleukin-6 (IL-6), IL-10, C-reactive protein (CRP) and serum amyloid A (SAA), at all sampling times in all patients.

RESULTS

The median (interquartile range) age in the HIFU and TCAP groups was 70 (62-77) and 68 (59-75) years, respectively (P = 0.4). Baseline levels (T0) of TNF-alpha, IL-6, IL-10, CRP and SAA were comparable in both groups. The levels of all four markers increased during both procedures; the median values were higher for TCAP at T2 (P = 0.02, 0.001, <0.001 and 0.06), T3 (P < 0.001, 0.009, 0.003 and 0.001), T4 (P = 0.007, <0.001, 0.005 and <0.001) and T5 (P < 0.001, 0.004, <0.001 and 0.02), respectively. IL-6 was also higher for TCAP at T1 (P = 0.03). IL-10 did not change at the different sampling times.

CONCLUSION

The tissue trauma and associated invasiveness of HIFU is less than that of TCAP, based on the variables objectively measured in this study.