Response: Design and reporting of clinical trials in LCH

Response:

Our collaborative research consistently seeks to identify interventions to improve the lives and outcome of children with Langerhans cell histiocytosis (LCH). The title and conclusion of the abstract of our paper,¹  questioned by Bernard et al in this issue, reflect 2 central messages: (1) there are no statistically significant differences in outcomes between the 2 intensive arms of LCH-II, but (2) there is a statistically significant superiority over the less intensive therapy of LCH-I (P < .001, considering all patients) with respect to the primary endpoint, the 6-week response.

Finding no evidence that one form of intensification was better than another in LCH-II, we conducted hypothesis-generating subgroup analysis of high-risk (RO+) patients to determine whether they responded differently from the less seriously affected patients without risk organ involvement (RO−) and found that they did. Contrary to prior evidence that age younger than 2 years is a separate risk factor for mortality, in LCH-II we reported the important finding of no deaths, in either arm, among such young RO− children.

We presented the combined analysis of arms of LCH-II versus LCH-I because we considered it unethical to bury the strong evidence of benefit with therapeutic intensification under the conclusions that there is no difference between arms within each protocol. We felt justified in this because both studies were randomized trials by the same investigators, targeting the same patient populations and differing only by the intensification of the chemotherapy. They were analogous (except for time) to arms of a unified trial design, thus relatively free of the problems usually encountered when drawing inferences across studies. In short, the benefit of the comparison outweighed the risks.

It would have been preferable from the point of study design to include an arm of LCH-I in the LCH-II protocol. However, 2 pieces of evidence made this problematic from an ethical point of view. The first was the disappointing lack of improvement in survival found in LCH-I. The second was that the earlier controlled, but not randomized, DAL trial²  suggested that more intensive therapy led to substantial improvement in outcomes.

Regarding some specific misconceptions raised in the letter of Bernard et al: (1) There is no standard treatment at this point for high-risk multisystem LCH. The goal of our randomized studies is to change this. (2) We never used the phrase “conventional and intensified treatment”; neither arm of LCH-II would have been considered a conventional treatment. (3) Therapy intensity is not defined by its clinical effect, as Bernard et al suggest, but by its composition. (4) Reference 4 cited in the letter of Bernard et al recommends conducting systematic reviews of randomized controlled trials. There are only 2 such trials in LCH. They are LCH-I and LCH-II, the subject of our paper. We believe that detailed analysis of the combined data from our 2 highly comparable controlled trials is far superior to a meta-analysis. (5) Reference 6 cited by Bernard et al pleads for exactly what we have done: registration and provision of open access to all Histiocyte Society clinical trials (LCH-I and LCH-II) before they were instituted, and full reporting of the results.

We appreciate the comment that the quality of the LCH-II trial is not in question. Our trials are developed with the collaboration of the membership of the Histiocyte Society. All collaborators have opportunities to have input into the study design. Although there is not always unanimity on all issues, all points of view are taken into account in making decisions in an effort to balance science with ethical treatment of histiocytosis patients. It is time to leave behind concerns already heard and answered in more than one venue, and to move forward to address important unsolved problems, such as reducing the risks of disease reactivation and permanent consequences, that will be answered best by forthcoming randomized trials.