1 Introduction
Dear Editor:
This Letter to the Editor is a secondary analysis of the data on a machine-learning algorithm for a new multivariate genetic panel that prognosticates the misuse of opioid prescribed for pain and subsequent development of Opioid Use Disorders.
The initiation pattern for the Opioid Use Disorders (OUD) has changed over the past five decades from the use of heroin in the community to medical prescriptions [1]. Prescription misuse of opioids (PMO) is seen in about 21–29% of patients with chronic pain [2]. The Global Burden of Disease Study showed a 22.3% increase in global opioid use disorder-related disability-adjusted life years between 2005 and 2015 [3]. To reduce this burden it is important to identify those patients with a high and low risk of getting initiated in to OUD with opioid prescriptions.
A machine-learning algorithm for a new multivariate genetic panel (MVGP) has undergone analytical (with derivation set) and cross (with separate internal data set) validation as a laboratory test to identifying patients at risk of developing OUD while on opioid prescriptions. This genetic panel in its derivation phase reached a sensitivity=82% and specificity=75%; training phase attained a sensitivity=92% and specificity=90%; generalizability of data phase achieved a sensitivity of 97% (95% CI: 90–100), specificity=87% with a positive likelihood ratio=7.3 (95% CI: 4.0–13.5), and negative likelihood ratio=0.03 (95% CI: 0.01–0.13) [4]. The post-test probability of MVGP using Bayesian approach was studied, which could improve the test interpretation, for the identification of those at risk for developing OUD while on prescriptions.
2 Methods
The positive likelihood ratio of 7.3 and negative likelihood ratio of 0.03 from the generalizability phase was used to establish the positive post-test probabilities (PPTP) and negative post-test probabilities (NPTP) of MVGP with the Fagan’s Nomogram for a pre-test probability (prevalence) of 29% [2]. We calculated the Clinical Utility Index [5] as a qualitative proxy for the clinical usefulness of the biomarker. STATA (version 15) statistical software was used for analyses.
3 Results
For LRP and LRN of 7.3 and 0.03, respectively, post-test probability for MVGP was 75% (95% CI=64, 83%; qualitative index=good) and negative post-test probability as 1% (95% CI=0, 8%; qualitative index=excellent) (Fig. 1).
The post-test probability of a multivariate genetic panel for predicting development Opioid Use Disorders on prescription medicines.
MVGP=multivariate genetic panel; PreProb=pretest-probability; LRP=positive likelihood ratio; LRN=negative likelihood ratio; PostProb=posttest-probability.
4 Discussion
The MVGP is a useful laboratory test that prognosticates the misuse of opioid prescribed for pain and later develop Opioid Use Disorders. If a patient requiring opioid prescription tests positive for MVGP, the chance of developing OUD increases from 29% to 75% (a clinically significant incremental increase in PPTP of 46%); the clinician can decide not to prescribe opioids. Contrariwise, if the patient tests negative the chance of not developing OUD decreases from 29% to 1% (a clinically significant incremental decrease in NPTP of 28%) and the patient can be prescribed opioids. This has been supported by the qualitative indices for clinical utility as well.
The major advantage of this biomarker is its ability to predict progression to OUD at the time of having to start an opioid prescription, for pain, thus prevention of developing OUD. Furthermore, screening of those who are MVGP positive during their treatment with opioids with simple psychological measures will aid in the case-finding of OUD. This clinical utility study is the logical next step in the translation of the identified biomarker from the derivative, and training stage in the laboratory to clinical use stage using the Baysean theorem. Despite the MVGP promising to be a useful predictive laboratory test for OUD while on prescriptions, more research is required from multi-omic and economic perspective in future.
-
Authors’s statements
-
Research funding: No funding involved.
-
Conflict of interest: The author has no conflicts of interest.
-
Informed consent: Secondary research, informed consent not required.
-
Ethical approval: Secondary research, ethical clearance not required.
References
[1] Cicero TJ, Ellis MS, Surratt HL, Kurtz SP. The changing face of heroin use in the United States: a retrospective analysis of the past 50 years. JAMA Psychiatry 2014;71:821–6.10.1001/jamapsychiatry.2014.366Search in Google Scholar PubMed
[2] Vowles KE, McEntee ML, Julnes PS, Frohe T, Ney JP, van der Goes DN. Rates of opioid misuse, abuse, and addiction in chronic pain: a systematic review and data synthesis. Pain 2015;156:569–76.10.1097/01.j.pain.0000460357.01998.f1Search in Google Scholar PubMed
[3] GBD 2015 DALYs and HALE Collaborators. Global, regional, and national disability-adjusted life-years (DALYs) for 315 diseases and injuries and healthy life expectancy (HALE), 1990–2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388:1603–58.10.1016/S0140-6736(16)31460-XSearch in Google Scholar PubMed PubMed Central
[4] Donaldson K, Demers L, Taylor K, Lopez J, Chang S. Multi-variant genetic panel for genetic risk of opioid addiction. Ann Clin Lab Sci 2017;47:452–6.Search in Google Scholar
[5] Mitchell AJ. Sensitivity×PPV is a recognized test called the clinical utility index (CUI+). Eur J Epidemiol 2011;26:251–2.10.1007/s10654-011-9561-xSearch in Google Scholar PubMed
©2020 Scandinavian Association for the Study of Pain. Published by Walter de Gruyter GmbH, Berlin/Boston. All rights reserved.
