Medical Decision Analysis: Choosing Second Line Treatment in Adults with Immune Thrombocytopenia (ITP)

Introduction

ITP is an autoimmune bleeding disorder characterized by accelerated destruction and reduced production of platelets. Steroids +/- IVIG are first line treatment, but there is no clear consensus on second line therapy. No randomized trials are available comparing the 3 most commonly considered second line options: thrombopoetin receptor agonists (TPO-RA), rituximab (Ritux) and splenectomy (SPL).This study used medical decision analysis, a mathematical Markov-model-based method, to compare second line treatments in adults with newly-diagnosed and persistent ITP.

Methods

Base case analysis considered 40-year old women with ITP lasting for at least 1 month, in daily need of corticosteroid treatment to maintain platelet counts above 30 x 10⁹ /L. We modeled three treatment strategies: SPL; TPO-RAs, and Ritux (with dexamethasone and mycophenolate mofetil). Patients who were unresponsive or relapsed after initial therapy received alternate treatment. Six different treatment strategies were evaluated (Table 1).

We simulated the course of identical patients using Markov processes which enabled complex time-dependent and recurrent clinical events, including side effects of treatments, to be modelled. At monthly intervals, patients could move from one clinical state to another based on probabilities assigned to these transitions. To facilitate estimating probabilities of transitioning between states, the cohort of patients was first divided into three categories according to the possible responsiveness to each treatment: i) responsive and curable- platelet count would increase above 50 x 10⁹/L after the treatment and remain there without further treatment, ii) responsive but incurable - platelet count would increase above 50 x 10⁹/L with treatment, but decrease below 30 x 10⁹/L if treatment was not continued, iii) non-responsive - platelet count would not increase above 30 x 10⁹/L despite treatment. This yielded 27 groups of patients for each treatment strategy since there are 3 possible groups for each of 3 strategies (3 x 3 x 3). Patients were followed for their entire lifetime.

Probability values and costs were collected from published data and expert opinions.

Within each monthly cycle of the Markov process, patients could fail the current treatment and switch to another. Complications - bleeding, thrombosis, or fever/sepsis - either minor or fatal, could occur. Sensitivity analysis was done to evaluate the accuracy of the input parameters and to determine for ranges of variables whether the outcomes were substantially affected.

Results

The results of estimated costs and years of lifespan for each treatment are presented in Table 2.

Discussion

This medical decision analysis fills the gap created by the absence of comparative second line treatment studies. In our model, patients suffering treatment-related toxicity or failing one treatment, attempt a second treatment and, failing that, progress to the third treatment. As a result, all treatable patients would eventually receive appropriate treatment. That explains why the lifespan with each of these 3 treatments in any order was within one year of each other at approximately 81 years of age. The order resulting in the longest lifespan was T-D-S which was driven by delay of SPL since SPL had the highest mortality composed of small contributions from the procedure itself, risk of thrombosis, and post-splenectomy sepsis. The D-T-S strategy, also delaying SPL, had almost the same lifespan at a significant cost savings. The lowest estimated cost was the S-D-T arm in which costs of additional therapy were reduced by doing SPL first and using TPO agents, if necessary, last. Of note, the cost estimates were based on costs of treatments obtained in the United States. Costs in Europe were estimated and were so much lower that the arms were essentially even.

The major limitation to our study is that quality of life was not included.

Conclusion: In the USA, TDS has the longest lifespan at a cost of approximately $130,000 per additional year of life. In Europe the cost for the additional year of life is < $20,000. If USA costs were as low as they are in Europe, there would be no compelling economic reason to select one of the specific second line therapies so treatment could be individualized according to physician and patient preference.

View largeDownload slide

View largeDownload slide

Close modal