Inconclusive Data in IA: Experts Fill the Void To Avoid Horror Vacui

ia-blog-header

Often we come across a study which raises more questions than it answers.  Sometimes we read a study publication again and again, and still we cannot figure out what the take-away learning is, as results are simply inconclusive or discordant. Occasionally we encounter studies with results that are so counterintuitive that even the authors have had difficulty putting it all together.  Then the interpretation becomes strangely warped, inventive or even more likely biased.

Such was the case with the study on the value of combination treatment for invasive aspergillosis published by Marr et al. last year [1].  It was an issue that still screams for clarification as the literature on echinocandin activity against aspergillosis is confusing to say the least.  The subject matter is important enough to deserve a definitive answer in a well-designed study.

The primary efficacy variable in this well executed, placebo controlled, double blind multicenter study was all-cause mortality by Week 6 in the modified ITT population.  Patients in the mITT population had proven or probable IA as was determined by blinded DRC case review. The study had a somewhat unusual design wherein voriconazole was given for 6 weeks but partnered with anidulafungin only for the first 2 weeks.

Patient Enrollment Dynamics: 93 study sites, 24 countries, 459 patients enrolled, 277 patients valid for efficacy, study duration: 3 years – an amazing feat!

It took the MSG 3 years and 93 study sites in 24 countries to collect 459 patients of which only 277 qualified for mITT analysis. Indeed, the study had to be extended for enrollment to accrue the required number of 250 mITT patients. This was clearly a herculean effort, certainly more difficult than anticipated.  It must have cost Pfizer, the sponsor of the study, a pretty penny.  Not to have worried, they would also reap the benefits as both drugs were from Pfizer.

As it turned out, all-cause mortality for the mITT population was 27.8% with mono-therapy, and 19.5% for combination therapy (delta: -8.3; CI: -19.0, 1.5); the P-value was 0.086. Hence, one cannot refute the Null-hypothesis of no difference between the treatment arms.

A secondary endpoint was Global Response, a measure of efficacy used in former studies of IA which correlates well with mortality.  This traditional outcome parameter allows us to create a link to prior trials. Well, in the Marr study, Global Response of success occurred with a higher frequency in patients receiving monotherapy.  Here the Table from the publication (Figure 1):

outcome-table
Figure 1: DRC adjudicated outcomes in the mITT population

Looking at the Kaplan-Meier plot (Figure 2), survival curves for mono- and combo-therapy hardly diverge before week 5½; in fact, they are almost superimposable. Are we supposed to think that anidulafungin, 4 weeks after dosing has ended, somehow magically boosts survival?  Such an interpretation defies biology.

The authors then do what all good investigators do: They performed a multi-variate analysis searching for factors that may have impacted outcome. Karnofsky score and the level of GM antigen at baseline were both strong predictors of mortality; hazard ratios indicate that combination therapy is better than mono-therapy when Karnofsky scores are low and entry GM antigen high.

k-m-graph
Figure 2: Kaplan-Meier plot – Mortality by Treatment Group

At this stage you are to be forgiven if you can’t figure out what to make of the study. Experts to the rescue: The authors kindly substitute what the data should have clarified with their belief system and tell us that 2 drugs are better than 1.  Their conclusion stated in the abstract and article is that “the combination of voriconazole and anidulafungin was associated with a nonsignificant but clinically meaningful survival benefit”, that “combination therapy led to higher survival in subgroups of patients with IA“.

Looking at the numbers, one is really baffled by such statements. It just seems nonsensical to conclude something not supported by the key study findings. Since when does data dredging trump prospectively defined endpoints?

We don’t like a toss-up situation either and do understand the frustration. The authors, siding with the sponsor, tell us:  When in doubt, believe in antifungals and combination therapy;  more just has to be better!

Well, not so, gents! We don’t follow the piper here. Your belief system is not necessarily ours.

Abbreviations:
IA          invasive aspergillosis
RCT       randomized controlled trial
GM        galactomannan
MSG      Mycoses Study Group

Reference:
[1] K Marr.  Combination Antifungal Therapy for Invasive Aspergillosis.  Ann Intern Med. 2015;162:81

Your Name (required)

Your Email (required)

Subject

Your Message

Print Friendly, PDF & Email

Leave a Reply

Your email address will not be published.