Validation of statistical predictive models meant to select melanoma patients for sentinel lymph node biopsy

Evaluation of the Indications for Sentinel Node Biopsy in Early-Stage Melanoma with the Advent of Adjuvant Systemic Therapy: An International, Multicenter Study

BACKGROUND

Patients presenting with early-stage melanoma (AJCC pT1b-pT2a) reportedly have a relatively low risk of a positive SNB (~5-10%). Those patients are usually found to have low-volume metastatic disease after SNB, typically reclassified to AJCC stage IIIA, with an excellent prognosis of ~90% 5-year survival. Currently, adjuvant systemic therapy is not routinely recommended for most patients with AJCC stage IIIA melanoma. The purpose was to assess the SN-positivity rate in early-stage melanoma and to identify primary tumor characteristics associated with high-risk nodal disease eligible for adjuvant systemic therapy METHODS: An international, multicenter retrospective cohort study from 7 large-volume cancer centers identified 3,610 patients with early primary cutaneous melanomas 0.8-2.0 mm in Breslow thickness (pT1b-pT2a; AJCC 8th edition). Patient demographics, primary tumor characteristics, and SNB status/details were analyzed.

RESULTS

The overall SNB-positivity rate was 11.4% (412/3610). Virtually all SNB-positive patients (409/412; 99.3%) were reclassified to AJCC stage IIIA. Multivariate analysis identified age, T-stage, mitotic rate, primary site and subtype, and lymphovascular invasion as independent predictors of sentinel node status. A mitotic rate of >1/mm2 was associated with a significantly increased SN-positivity rate and was the only significant independent predictor of high-risk SNB metastases (>1 mm maximum diameter).

CONCLUSIONS

The new treatment paradigm brings into question the role of SNB for patients with early-stage melanoma. The results of this large international cohort study suggest that a reevaluation of the indications for SNB for some patients with early-stage melanoma is required.

MelRisk: Using neutrophil-to-lymphocyte ratio to improve risk prediction models for metastatic cutaneous melanoma in the sentinel lymph node

BACKGROUND

Identifying metastatic melanoma in the sentinel lymph node (SLN) is important because 80% of SLN biopsies are negative and 11% of patients develop complications. The neutrophil-to-lymphocyte ratio (NLR), a biomarker of micrometastatic disease, could improve prediction models for SLN status. We externally validated existing models and developed 'MelRisk' prognostic score to better predict SLN metastasis.

METHODS

The models were externally validated using data from a multicenter cohort study of 1,251 adults. Additionally, we developed and internally validated a new prognostic score `MelRisk', using candidate predictors derived from the extant literature.

RESULTS

The Karakousis model had a C-statistic of 0.58 (95% CI, 0.54-0.62). The Sondak model had a C-statistic of 0.57 (95% CI 0.53-0.61). The MIA model had a C-statistic of 0.60 (95% CI. 0.56-0.64). Our 'MelRisk' model (which used Breslow thickness, ulceration, age, anatomical site, and the NLR) showed an adjusted C-statistic of 0.63 (95% CI, 0.56-0.64).

CONCLUSION

Our prediction tool is freely available in the Google Play Store and Apple App Store, and we invite colleagues to externally validate its performance .

Machine learning directed sentinel lymph node biopsy in cutaneous head and neck melanoma

BACKGROUND

The specificity of sentinel lymph node biopsy (SLNB) for detecting lymph node metastasis in head and neck melanoma (HNM) is low under current National Comprehensive Cancer Network (NCCN) treatment guidelines.

METHODS

Multiple machine learning (ML) algorithms were developed to identify HNM patients at very low risk of occult nodal metastasis using National Cancer Database (NCDB) data from 8466 clinically node negative HNM patients who underwent SLNB. SLNB performance under NCCN guidelines and ML algorithm recommendations was compared on independent test data from the NCDB (n = 2117) and an academic medical center (n = 96).

RESULTS

The top-performing ML algorithm (AUC = 0.734) recommendations obtained significantly higher specificity compared to the NCCN guidelines in both internal (25.8% vs. 11.3%, p < 0.001) and external test populations (30.1% vs. 7.1%, p < 0.001), while achieving sensitivity >97%.

CONCLUSION

Machine learning can identify clinically node negative HNM patients at very low risk of nodal metastasis, who may not benefit from SLNB.

The neutrophil-lymphocyte ratio and locoregional melanoma: a multicentre cohort study

OBJECTIVES

The neutrophil-lymphocyte ratio (NLR) is an inflammatory biomarker which is useful in cancer prognostication. We aimed to investigate the differences in baseline NLR between patients with localised and metastatic cutaneous melanoma and how this biomarker changed over time with the recurrence of disease.

METHODS

This multicentre cohort study describes patients treated for Stage I-III cutaneous melanoma over 10 years. The baseline NLR was measured immediately prior to surgery and again at the time of discharge or disease recurrence. The odds ratios (OR) for sentinel node involvement are estimated using mixed-effects logistic regression. The risk of recurrence is estimated using multivariable Cox regression.

RESULTS

Overall 1489 individuals were included. The mean baseline NLR was higher in patients with palpable nodal disease compared to those with microscopic nodal or localised disease (2.8 versus 2.4 and 2.3, respectively; p < 0.001). A baseline NLR ≥ 2.3 was associated with 30% higher odds of microscopic metastatic melanoma in the sentinel lymph node [adjusted OR 1.3 (95% CI 1.3, 1.3)]. Following surgery, 253 patients (18.7%) developed recurrent melanoma during surveillance although there was no statistically significant association between the baseline NLR and the risk of recurrence [adjusted HR 0.9 (0.7, 1.1)].

CONCLUSION

The NLR is associated with the volume of melanoma at presentation and may predict occult sentinel lymph metastases. Further prospective work is required to investigate how NLR may be modelled against other clinicopathological variables to predict outcomes and to understand the temporal changes in NLR following surgery for melanoma.

Locally Weighted Score Estimation for Quantile Classification in Binary Regression Models

One common use of binary response regression methods is classification based on an arbitrary probability threshold dictated by the particular application. Since this is given to us a priori, it is sensible to incorporate the threshold into our estimation procedure. Specifically, for the linear logistic model, we solve a set of locally weighted score equations, using a kernel-like weight function centered at the threshold. The bandwidth for the weight function is selected by cross validation of a novel hybrid loss function that combines classification error and a continuous measure of divergence between observed and fitted values; other possible cross-validation functions based on more common binary classification metrics are also examined. This work has much in common with robust estimation, but diers from previous approaches in this area in its focus on prediction, specifically classification into high- and low-risk groups. Simulation results are given showing the reduction in error rates that can be obtained with this method when compared with maximum likelihood estimation, especially under certain forms of model misspecification. Analysis of a melanoma data set is presented to illustrate the use of the method in practice.

Impact of Completion Lymph Node Dissection on Patients with Positive Sentinel Lymph Node Biopsy in Melanoma

BACKGROUND

Whether patients with positive SLNB should undergo complete lymph node dissection (CLND) is an important unanswered clinical question.

STUDY DESIGN

Patients diagnosed with positive SLNB at a melanoma referral center from 1991 to 2013 were studied. Outcomes of patients who underwent CLND were compared with those who did not undergo immediate CLND (observation [OBS] group).

RESULTS

There were 471 patients who had positive SLNB; 375 (79.6%) in the CLND group and 96 (20.4%) in the OBS group. The groups were similar except that the CLND group was younger and had more sentinel nodes removed. Five-year nodal recurrence-free survival was significantly better in the CLND group compared with the OBS group (93.1% vs 84.4%; p = 0.005). However, 5-year (66.4% vs 55.2%) and 10-year (59.5% vs 45.0%) distant metastasis-free survival rates were not significantly different (p = 0.061). The CLND group's melanoma-specific survival (MSS) rate was superior to that of the OBS group; 5-year MSS rates were 73.7% vs 65.5% and 10-year MSS rates were 66.8% vs 48.3% (p = 0.015). On multivariate analysis, CLND was associated with improved MSS (hazard ratio = 0.60; 95% CI, 0.40-0.89; p = 0.011) and lower nodal recurrence (hazard ratio = 0.46; 95% CI, 0.24-0.86; p = 0.016). Increased Breslow thickness, older age, ulceration, and trunk melanoma were all associated with worse outcomes. On subgroup analysis, the following factors were associated with better outcomes from CLND: male sex, nonulcerated primary, intermediate thickness, Clark level IV or lower extremity tumors.

CONCLUSIONS

Treatment of positive SLNB with CLND was associated with improved MSS and nodal recurrence rates. Follow-up beyond 5 years was needed to see a significant difference in MSS rates.

Tumor Cell Adhesion As a Risk Factor for Sentinel Lymph Node Metastasis in Primary Cutaneous Melanoma

PURPOSE

Less than 20% of patients with melanoma who undergo sentinel lymph node (SLN) biopsy based on American Society of Clinical Oncology/Society of Surgical Oncology recommendations are SLN positive. We present a multi-institutional study to discover new molecular risk factors associated with SLN positivity in thin and intermediate-thickness melanoma.

PATIENTS AND METHODS

Gene clusters with functional roles in melanoma metastasis were discovered by next-generation sequencing and validated by quantitative polymerase chain reaction using a discovery set of 73 benign nevi, 76 primary cutaneous melanoma, and 11 in-transit melanoma metastases. We then used polymerase chain reaction to quantify gene expression in a model development cohort of 360 consecutive thin and intermediate-thickness melanomas and a validation cohort of 146 melanomas. Outcome of interest was SLN biopsy metastasis within 90 days of melanoma diagnosis. Logic and logistic regression analyses were used to develop a model for the likelihood of SLN metastasis from molecular, clinical, and histologic variables.

RESULTS

ITGB3, LAMB1, PLAT, and TP53 expression were associated with SLN metastasis. The predictive ability of a model that included these molecular variables in combination with clinicopathologic variables (patient age, Breslow depth, and tumor ulceration) was significantly greater than a model that only considered clinicopathologic variables and also performed well in the validation cohort (area under the curve, 0.93; 95% CI, 0.87 to 0.97; false-positive and false-negative rates of 22% and 0%, respectively, using a 10% cutoff for predicted SLN metastasis risk).

CONCLUSION

The addition of cell adhesion-linked gene expression variables to clinicopathologic variables improves the identification of patients with SLN metastases within 90 days of melanoma diagnosis.

Validation of a nomogram predicting sentinel lymph node status in melanoma in an Irish population

BACKGROUND

Sentinel lymph node (SLN) positivity is an important prognostic factor in cutaneous melanoma. A nomogram has been developed at Memorial Sloan-Kettering Cancer Centre (MSKCC) to predict SLN positivity and this may be useful to select patients for SLN biopsy.

AIMS

We aimed to determine whether this nomogram would be of clinical use in an Irish population.

METHODS

Age, Breslow thickness, Clark's level, presence of ulceration and tumour location indices were used to calculate the probability of SLN positivity with the MSKCC nomogram in 124 patients who underwent SLN biopsy in Beaumont Hospital between 2006 and 2012. Discrimination and calibration of the nomogram were evaluated. Negative predictive value (NPV) of the nomogram was calculated, using a cut-off of nomogram predicted probability of <9%.

RESULTS

SLN biopsy was positive in 25 patients (20.16%). Overall predictive accuracy of the nomogram was found to be significant with an area under the curve of 0.805 (95% confidence interval 0.710-0.899). The mean predicted probability correlated well with observed risk (r = 0.887). The NPV was 92.86% with an error rate of 3.23%. This would lead to a reduction in SLN biopsy rate of 45.16%.

CONCLUSIONS

This nomogram is valid and accurate at predicting SLN positivity in an Irish population. This may facilitate the clinical decision to perform a SLN biopsy in malignant melanoma.

Clinical impact of biopsy method on the quality of surgical management in melanoma

BACKGROUND AND OBJECTIVES

Though guidelines recommend excisional biopsy for diagnosing melanoma, partial biopsy techniques are commonly performed, risking underestimation of Breslow depth and altering surgical management. Biopsy choice patterns by specialty and subsequent impact on surgical management was examined.

METHODS

Retrospective review of the University of Vermont Cancer Registry. All patients with a single, primary cutaneous melanoma from 1/02 to 6/12 analyzed.

RESULTS

Among 853 lesions analyzed, 606 had Breslow depth <1 mm. Dermatologists perform 62.6% of biopsies, favoring shave biopsies; surgeons favor excisional biopsies (48%), and primary care doctors favor punch biopsies (44.8%), (P < 0.001). Final Breslow depth was upstaged in 107 (12.5%); however, only 23 of 488 partial biopsies (4.7%) displayed a discrepancy great enough to change surgical recommendations (P < 0.001). There was no statistically significant relationship with presence of ulceration, regression, high Clark level, or high mitotic index.

CONCLUSIONS

Partial biopsy techniques are commonly performed in diagnosing melanoma; especially among dermatologists, who perform the majority of biopsies. Though partial biopsies were less accurate in determining Breslow thickness; they rarely alter recommendations for surgical management. Predictive features could not be determined to identify the few cases where a Breslow discrepancy was clinically relevant.

Sentinel lymph node biopsy for melanoma: American Society of Clinical Oncology and Society of Surgical Oncology joint clinical practice guideline

PURPOSE

The American Society of Clinical Oncology (ASCO) and Society of Surgical Oncology (SSO) sought to provide an evidence-based guideline on the use of lymphatic mapping and sentinel lymph node (SLN) biopsy in staging patients with newly diagnosed melanoma.

METHODS

A comprehensive systematic review of the literature published from January 1990 through August 2011 was completed using MEDLINE and EMBASE. Abstracts from ASCO and SSO annual meetings were included in the evidence review. An Expert Panel was convened to review the evidence and develop guideline recommendations.

RESULTS

Seventy-three studies met full eligibility criteria. The evidence review demonstrated that SLN biopsy is an acceptable method for lymph node staging of most patients with newly diagnosed melanoma.

RECOMMENDATIONS

SLN biopsy is recommended for patients with intermediate-thickness melanomas (Breslow thickness, 1 to 4 mm) of any anatomic site; use of SLN biopsy in this population provides accurate staging. Although there are few studies focusing on patients with thick melanomas (T4; Breslow thickness, > 4 mm), SLN biopsy may be recommended for staging purposes and to facilitate regional disease control. There is insufficient evidence to support routine SLN biopsy for patients with thin melanomas (T1; Breslow thickness, < 1 mm), although it may be considered in selected patients with high-risk features when staging benefits outweigh risks of the procedure. Completion lymph node dissection (CLND) is recommended for all patients with a positive SLN biopsy and achieves good regional disease control. Whether CLND after a positive SLN biopsy improves survival is the subject of the ongoing Multicenter Selective Lymphadenectomy Trial II.