Support Vector Machine Learning Model for the Prediction of Sentinel Node Status in Patients With Cutaneous Melanoma

Recent Advances in the Field of Artificial Intelligence for Precision Medicine in Patients with a Diagnosis of Metastatic Cutaneous Melanoma

Standard-of-care medical imaging techniques such as CT, MRI, and PET play a critical role in managing patients diagnosed with metastatic cutaneous melanoma. Advancements in artificial intelligence (AI) techniques, such as radiomics, machine learning, and deep learning, could revolutionize the use of medical imaging by enhancing individualized image-guided precision medicine approaches. In the present article, we will decipher how AI/radiomics could mine information from medical images, such as tumor volume, heterogeneity, and shape, to provide insights into cancer biology that can be leveraged by clinicians to improve patient care both in the clinic and in clinical trials. More specifically, we will detail the potential role of AI in enhancing detection/diagnosis, staging, treatment planning, treatment delivery, response assessment, treatment toxicity assessment, and monitoring of patients diagnosed with metastatic cutaneous melanoma. Finally, we will explore how these proof-of-concept results can be translated from bench to bedside by describing how the implementation of AI techniques can be standardized for routine adoption in clinical settings worldwide to predict outcomes with great accuracy, reproducibility, and generalizability in patients diagnosed with metastatic cutaneous melanoma.

Artificial intelligence in melanoma: A systematic review

BACKGROUND

Melanoma accounts for the majority of skin cancer deaths. Artificial intelligence has been applied in many types of cancers, and in melanoma in recent years. However, no systematic review summarized the application of artificial intelligence in melanoma.

AIMS

This study aims to systematically review previously published articles to explore the application of artificial intelligence in melanoma.

MATERIALS & METHODS

PubMed database was used to search the eligible publications on August 1, 2020. The query term was "artificial intelligence" and "melanoma."

RESULTS

A total of 51 articles were included in this review. Artificial intelligence technique is mainly used in the evaluation of dermoscopic images, other image segmentation and processing, and artificial intelligence diagnosis system.

DISCUSSION

Artificial intelligence is also applied in metastasis prediction, drug response prediction, and prognosis of melanoma. Besides, patients' perspectives of artificial intelligence and collaboration of human and artificial intelligence in melanoma also attracted attention. The query term might not include all articles, and we could not examine the algorithms that were built without publication.

CONCLUSION

The performance of artificial intelligence in melanoma is satisfactory and the future for potential applications is enormous.

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.

The prognostic significance of microsatellites in cutaneous melanoma

Microscopic satellite metastases are an adverse prognostic feature in primary cutaneous melanoma patients. The prognostic significance of microsatellites, including their number, size and distance from the primary melanoma, using the 8th edition American Joint Committee on Cancer definition, has not previously been evaluated. This study sought to determine the prognostic significance of microsatellites in histopathologically reviewed cases. Eighty-seven cases of primary cutaneous melanoma with the presence of microsatellites documented in the original pathology report and all histopathology slides available were reviewed and the findings were correlated with clinical outcome. Matched control cases were selected for all confirmed microsatellites cases. The presence of microsatellites was confirmed in 69 cases. The microsatellite group had significantly worse prognosis, with 21% 5-year disease-free survival compared with 56% in the control group (p < 0.001). The 5-year melanoma-specific survival was 53% in the microsatellites group and 73% in the control group (p = 0.004). Increasing distance (mm) of the microsatellite from the primary melanoma was found to adversely influence disease-free survival (HR = 1.24, 95% CI: 1.13-1.36, p < 0.001), overall survival (HR = 1.26 95%CI: 1.13-1.40, p < 0.001), and melanoma-specific survival (HR = 1.27 95% CI: 1.11-1.45, p < 0.001). Number and size of microsatellites were not significant prognostic factors. The presence of microsatellites was the only factor that proved to be an independent predictor of sentinel node positivity in multivariate analysis (OR 4.64; 95% CI 1.66-12.95; p = 0.003). Microsatellites were significantly associated with more loco-regional recurrences (p < 0.001) but not distant metastases (p = 0.821). Melanomas with microsatellites as defined by the 8th edition American Joint Committee on Cancer staging system are thus aggressive tumors, associated with significantly worse disease-free survival, overall survival and melanoma-specific survival. The presence of microsatellites is also associated with sentinel node-positivity and local and in-transit recurrence. Increasing distance of the microsatellite from the primary tumor is an independent adverse prognostic factor that warrants further evaluation.

Tumor mitotic rate added to the equation: melanoma prognostic factors changed? : a single-institution database study on the prognostic value of tumor mitotic rate for sentinel lymph node status and survival of cutaneous melanoma patients

BACKGROUND

This study aimed to investigate the predictive value of the tumor mitotic rate per mm(2) (TMR) for sentinel lymph node (SLN) status and survival in intermediate and thick cutaneous melanoma.

METHODS

Patients treated for stage I and II melanoma with wide local excision and SLN biopsy between May 1995 and May 2013 were analyzed. In case of insufficient data regarding TMR, pathology slides were reanalyzed. Prognostic factors for SLN status and survival were analyzed with the emphasis on TMR, which was analyzed as a continuous variable, dichotomized (median value) and categorized by two methods.

RESULTS

The study analyzed 453 patients with complete TMR data. The median Breslow thickness was 2.20 mm, and 31.8 % of patients had tumor-positive sentinel lymph node biopsies (SLNBs). In the univariate analysis, TMR was associated with tumor-positive SLNB. This association was not significant in the multivariate analysis. Breslow thickness, primary tumor location on trunk and legs, and younger age were associated with tumor-positive SNLB. At a median follow-up of 47 months, 119 patients (26.3 %) had recurrent disease, and 92 (20.3 %) had died of melanoma. In the univariate analysis, TMR could be established as a significant prognostic factor for disease-free and disease-specific survival, but not in the multivariate analyses. Breslow thickness, ulcerated melanoma, and tumor-positive SLNB were significant prognostic factors for survival.

CONCLUSION

The study was unable to establish TMR as an independent prognostic factor associated with the presence of SLN metastasis. Regarding survival, increasing TMR showed a strong association with decreased survival in the univariate analysis, but this association was rendered nonsignificant by the importance of Breslow thickness and ulceration status in the multivariate model.

Utility of propidium monoazide viability assay as a biomarker for a tuberculosis disease

Reliable laboratory diagnosis of tuberculosis (TB), including laboratory biomarkers of cure, remains a challenge. In our study we evaluated the performance of a Propidium Monoazide (PMA) assay for the detection of viable TB bacilli in sputum specimens during anti-TB chemotherapy and its potential use as a TB biomarker. The study was conducted at three centres on 1937 sputum specimens from 310 adult bacteriologically confirmed pulmonary TB patients obtained before commencing anti-TB treatment and at regular intervals afterwards. Performance of the PMA assay was assessed using various readout assays with bacteriology culture results and time to positivity on liquid media used as reference standards. Treatment of sputum with N-acetyl-cysteine was found to be fully compatible with the PMA assay. Good sensitivity and specificity (97.5% and 70.7-80.0%) for detection of live TB bacilli was achieved using the Xpert(®) MTB/RIF test as a readout assay. Tentative Ct and ΔCt thresholds for the Xpert(®) MTB/RIF system were proposed. Good correlation (r = 0.61) between Ct values and time to positivity of TB cultures on liquid media was demonstrated. The PMA method has potential in monitoring bacterial load in sputum specimens and so may have a role as a biomarker of cure in TB treatment.

Epithelial-mesenchymal transition biomarkers and support vector machine guided model in preoperatively predicting regional lymph node metastasis for rectal cancer

Background:

Current imaging modalities are inadequate in preoperatively predicting regional lymph node metastasis (RLNM) status in rectal cancer (RC). Here, we designed support vector machine (SVM) model to address this issue by integrating epithelial–mesenchymal-transition (EMT)-related biomarkers along with clinicopathological variables.

Methods:

Using tissue microarrays and immunohistochemistry, the EMT-related biomarkers expression was measured in 193 RC patients. Of which, 74 patients were assigned to the training set to select the robust variables for designing SVM model. The SVM model predictive value was validated in the testing set (119 patients).

Results:

In training set, eight variables, including six EMT-related biomarkers and two clinicopathological variables, were selected to devise SVM model. In testing set, we identified 63 patients with high risk to RLNM and 56 patients with low risk. The sensitivity, specificity and overall accuracy of SVM in predicting RLNM were 68.3%, 81.1% and 72.3%, respectively. Importantly, multivariate logistic regression analysis showed that SVM model was indeed an independent predictor of RLNM status (odds ratio, 11.536; 95% confidence interval, 4.113–32.361; P<0.0001).

Conclusion:

Our SVM-based model displayed moderately strong predictive power in defining the RLNM status in RC patients, providing an important approach to select RLNM high-risk subgroup for neoadjuvant chemoradiotherapy.

Sentinel node status prediction by four statistical models: results from a large bi-institutional series (n = 1132)

OBJECTIVE

To improve selection for sentinel node (SN) biopsy (SNB) in patients with cutaneous melanoma using statistical models predicting SN status.

SUMMARY BACKGROUND DATA

About 80% of patients currently undergoing SNB are node negative. In the absence of conclusive evidence of a SNBassociated survival benefit, these patients may be over-treated. Here, we tested the efficiency of 4 different models in predicting SN status.

METHODS

The clinicopathologic data (age, gender, tumor thickness, Clark level, regression, ulceration, histologic subtype, and mitotic index) of 1132 melanoma patients who had undergone SNB at institutions in Italy and Australia were analyzed. Logistic regression, classification tree, random forest, and support vector machine models were fitted to the data. The predictive models were built with the aim of maximizing the negative predictive value (NPV) and reducing the rate of SNB procedures though minimizing the error rate.

RESULTS

After cross-validation logistic regression, classification tree, random forest, and support vector machine predictive models obtained clinically relevant NPV (93.6%, 94.0%, 97.1%, and 93.0%, respectively), SNB reduction (27.5%, 29.8%, 18.2%, and 30.1%, respectively), and error rates (1.8%, 1.8%, 0.5%, and 2.1%, respectively).

DISCUSSION

Using commonly available clinicopathologic variables, predictive models can preoperatively identify a proportion of patients ( approximately 25%) who might be spared SNB, with an acceptable (1%-2%) error. If validated in large prospective series, these models might be implemented in the clinical setting for improved patient selection, which ultimately would lead to better quality of life for patients and optimization of resource allocation for the health care system.

Clinical considerations on sentinel node biopsy in melanoma from an Italian multicentric study on 1,313 patients (SOLISM-IMI)

BACKGROUND

Although widely used for the management of patients with cutaneous melanoma, the sentinel lymph node (SLN) biopsy (SNB) procedure raises several issues. This study was designed to investigate: the predictive factors of SLN status, the false-negative (FN) rate, and patients' prognosis after SNB.

PATIENTS AND METHODS

This is an observational, prospective study conducted on a large series of consecutive patients (n = 1,313) enrolled by 23 Italian centers from 2000 through 2002. A commonly shared protocol was adopted for the SNB surgical procedure and the SLN pathological examination.

RESULTS

The SLN positive and false-negative (FN) rates were 16.9% and 14.4%, respectively (median follow-up, 4.5 years). At multivariable logistic regression analysis, the frequency of positive SLN increased with increasing Breslow thickness (p < 0.0001) and decreased in patients with melanoma regression (p = 0.024). At the multivariable Cox regression analysis, SLN status was the most important prognostic factor (hazards ratio (HR) = 3.08) for overall survival; the other statistically significant factors were sex, age, Breslow thickness, and Clark's level. Considering SLN and NSLN status, including FN cases, we identified four groups of patients with different prognoses. The 5-year overall survival of patients with positive SLNs was 71.3% in those with negative nonsentinel lymph nodes (NSLNs) and 50.4% if NSLNs were positive.

CONCLUSIONS

Regression in the primary melanoma seems to be a protective factor from metastasis in the SLN. When correctly calculated, the SNB FN rate is 15-20%. Furthermore, the SNB is important to more precisely assess the prognosis of patients with melanoma.

Factors predictive of nonsentinel lymph node involvement and clinical outcome in melanoma patients with metastatic sentinel lymph node

BACKGROUND

Identification of melanoma patients who need completion lymphadenectomy and adjuvant treatment after positive sentinel lymph node (SLN) biopsy would be a fundamental step forward toward personalized medicine. This study tested the hypothesis that the microscopic features of metastatic SLNs might predict not only nonsentinel lymph node (NSLN) status, but also patients' clinical outcomes.

METHODS

A retrospective analysis was performed on 96 consecutive melanoma patients who underwent completion lymphadenectomy after positive SLN biopsy. Patients' age and sex, primary tumor Breslow thickness, number of positive SLNs, the largest diameter and depth of invasion of metastatic deposits in the SLN, S stage, and pattern of nodal involvement were correlated with the presence of metastatic disease in NSLNs as well as with the likelihood of tumor recurrence and patient death.

RESULTS

At pathological examination, 20 patients (20.8%) had metastatic melanoma in the NSLN. Pattern of nodal involvement, depth of invasion of SLN by metastatic disease, and S stage were statistically significantly associated with the presence of metastatic disease in NSLN. Multivariate analysis revealed that only the SLN depth of invasion was an independent predictor of NSLN status (P = .0035). This parameter was also significantly associated with disease-free and overall survival, both by univariate (P < .0001 and P = .0006, respectively) and multivariate (P < .0001 and P = .0013, respectively) survival analysis.

CONCLUSIONS

These findings support further investigation of SLN depth of invasion as a predictive factor of potential clinical use to select patients as candidates for completion lymphadenectomy and adjuvant treatment.

Effects of SVM parameter optimization on discrimination and calibration for post-procedural PCI mortality

Support vector machines (SVM) have become popular among machine learning researchers, but their applications in biomedicine have been somewhat limited. A number of methods, such as grid search and evolutionary algorithms, have been utilized to optimize model parameters of SVMs. The sensitivity of the results to changes in optimization methods has not been investigated in the context of medical applications. In this study, radial-basis kernel SVM and polynomial kernel SVM mortality prediction models for percutaneous coronary interventions were optimized using (a) mean-squared error, (b) mean cross-entropy error, (c) the area under the receiver operating characteristic, and (d) the Hosmer-Lemeshow goodness-of-fit test (HL chi(2)). A threefold cross-validation inner and outer loop method was used to select the best models using the training data, and evaluations were based on previously unseen test data. The results were compared to those produced by logistic regression models optimized using the same indices. The choice of optimization parameters had a significant impact on performance in both SVM kernel types.