Breakthrough Cancer Pain Clinical Features and Differential Opioids Response: A Machine Learning Approach in Patients With Cancer From the IOPS-MS Study

Abstract

Purpose: A large proportion of patients with cancer suffer from breakthrough cancer pain (BTcP). Several unmet clinical needs concerning BTcP treatment, such as optimal opioid dosages, are being investigated. In this analysis the hypothesis, we explore with an unsupervised learning algorithm whether distinct subtypes of BTcP exist and whether they can provide new insights into clinical practice.

Methods: Partitioning around a k-medoids algorithm on a large data set of patients with BTcP, previously collected by the Italian Oncologic Pain Survey group, was used to identify possible subgroups of BTcP. Resulting clusters were analyzed in terms of BTcP therapy satisfaction, clinical features, and use of basal pain and rapid-onset opioids. Opioid dosages were converted to a unique scale and the BTcP opioids-to-basal pain opioids ratio was calculated for each patient. We used polynomial logistic regression to catch nonlinear relationships between therapy satisfaction and opioid use.

Results: Our algorithm identified 12 distinct BTcP clusters. Optimal BTcP opioids-to-basal pain opioids ratios differed across the clusters, ranging from 15% to 50%. The majority of clusters were linked to a peculiar association of certain drugs with therapy satisfaction or dissatisfaction. A free online tool was created for new patients' cluster computation to validate these clusters in future studies and provide handy indications for personalized BTcP therapy.

Conclusion: This work proposes a classification for BTcP and identifies subgroups of patients with unique efficacy of different pain medications. This work supports the theory that the optimal dose of BTcP opioids depends on the dose of basal opioids and identifies novel values that are possibly useful for future trials. These results will allow us to target BTcP therapy on the basis of patient characteristics and to define a precision medicine strategy also for supportive care.

FIG 1.

(A) Algorithm used for the diagnosis of breakthrough cancer pain (BTcP) during patients’ enrollment in the Italian Oncologic Pain Survey (modified from Mercadante et al). (B) A two-dimensional t-distributed stochastic neighbor embedding projection of all patients, colored by their clusters, on the basis of the following BTcP features: number of BTcP episodes, BTcP peaks duration, BTcP type, BTcP intensity, number of days since the beginning of BTcP episodes, eventual benefit from pharmacotherapy, eventual benefit from rest, and whether BTcP was enhanced by movements. Each point represents a patient. Patients’ dissimilarity in BTcP clinical features is represented by the points distance. Colors represent 12 clusters computed through partitioning around the medoids (k-medoids) algorithm.

FIG 2.

Defining features of the 12 breakthrough cancer pain (BTcP) clusters. Plots represent in order: (A) BTcP intensity using numeric rating scale, (B) BTcP peak duration, (C) BTcP type, (D) number of BTcP events per day, (E) presence of benefit in BTcP management with pharmacotherapy, (F) presence of benefit in BTcP management with rest, (G) presence of BTcP activation with movements, and (H) days since BTcP episodes started.

FIG 3.

(A) Correlation of breakthrough cancer pain (BTcP) therapy satisfaction with BTcP opioid dose and basal pain opioid dose ratio. (B) BTcP opioid drug dose alone, and (C) basal pain opioid drug dose. Solid lines represent logistic regressions calculated with more than 1 degree of freedom and dashed lines represent 95% CIs. (D) Correlation between fast to basal opioids ratio and therapy satisfaction for each cluster. Exp(OR), exponent (odds ratio).

FIG A1.

(A) Silhouette statistics for clusters 2 to 30. The appropriate number of clusters for additional analyses was found to be 12, an optimal trade-off between the average width of clusters silhouette (0.45) and the interpretability of clusters themselves. (B) Heatmap showing the concordance of PAM clustering with complete-linkage hierarchical clustering. The Rand index, which compares the replicability of the two algorithms, was 0.89.