The use of antifungals in combination in the primary treatment of invasive aspergillosis has been increasing despite the fact that the available data supporting this practice is, at best, contradictory.1 This issue has now been addressed in a large clinical trial.

Marr and colleagues at 93 international sites randomly assigned 454 patients  the intent-to-treat (ITT) population  with underlying hematologic malignancy or hematopoietic stem cell transplantation with suspected or documented invasive aspergillosis (IA) to blindly receive either anidulafungin or placebo, each together with voriconazole. Voriconazole was administered intravenously for the first week, after which it could be given by mouth, while anidulafungin or placebo was administered intravenously for 2-4 weeks. The primary efficacy endpoint was mortality at 6 weeks in the 277 patients in whom the diagnosis of IA was confirmed to be either possible, probable, or proven  who comprised the modified intent-to-treat (mITT) population. Secondary pre-specified end points were 6- and 12-week mortality in 6 subgroups believed to have prognostic import.

In addition to consistent radiographic findings, the diagnosis was proven in only 1.4% of patients, and was probable based on a galactomannan index > 0.5 in serum or bronchoalveolar lavage (BAL) specimen in 77.5% and on positive histopathology, culture, and/or cytology in 21.1%. Combination treatment (voriconazole and anidulafungin or placebo) was administered for a median of 14 days (range, 1 to 29) while the median duration of voriconazole administration was 42 days (range, 1 to 48). Voriconazole serum concentrations did not differ between treatment groups in the subset in which it was measured.

Six-week mortality in the ITT population was 20.6% (47 of 228) in the combination therapy group and 23.5% (53 of 226) in those receiving monotherapy. In the mITT population, mortality rates at 6 weeks were 19.5% (26 of 135) and 27.8% (26 of 135), respectively (95% CI of the difference, -19.0 to 1.5; 2-siede P=0.087). A successful global response was achieved at 6 weeks in 44 (32.6%) of combination therapy patients and in 61 (43.0%) of those given monotherapy (95% CI for the difference, -21.6 to 1.2). The frequency of hepatobiliary adverse events was higher in the combination group than in those receiving monotherapy (12.7% vs. 8.4%).

A post hoc analysis identified low Karnofsky score, low platelet count, and high maximum baseline galactomannan as independent prognostic factors for 6-week mortality. Not having found their desired outcome, the investigators subsequently searched for apparent differences in outcome between subgroups in addition to the multiple ones they had pre-specified  and they found one. A further post hoc analysis examined outcomes in patients with elevated serum or BAL and found that the all-cause mortality was 15.7% (17 of 108) in the combination therapy group compared with 27.3% (30 of 110) in the monotherapy group (difference, −11.6 percentage points [CI, −22.7 to −0.4]; P = 0.037). This apparent difference, however, appears to be largely limited to individuals with galactomannan indices in the range of > 0.5–1.5, with no significant difference between treatment groups in those with higher indices.


One issue not addressed in this study that could have affected results is the emergence of resistance to voriconazole in Europe, where many of the study centers were located.

This ambitious study failed to identify a survival advantage from the addition of anidulafungin to voriconazole in the initial portion of primary treatment in selected patients with suspected or documented aspergillosis. On further digging into the results, however, they concluded:

• From the Discussion: “We conclude that treatment of IA with the combination of voriconazole and anidulafungin was associated with a nonsignificant but clinically meaningful survival benefit in patients with HM or HCT.”

 From the Abstract: “Compared with voriconazole monotherapy, combination therapy with anidulafungin led to higher survival in subgroups of patients with IA. Limitations in power preclude definitive conclusions about superiority.”

 From the Editors’ Notes: “Mortality was lower with combination therapy in a subgroup of patients whose IA diagnosis was established by radiographic findings and galactomannan positivity.”

Thus, despite the lack of support from the results of the predetermined primary outcome, the authors (and editors) make a strong pitch for combination therapy, based on an exploratory post hoc subset analysis that appears to apply only to individuals with a modest range of galactomannan elevation. The validity of such a subgroup analysis deserves examination.

Sun and colleagues have listed four criteria for the assessment of subgroup analyses within individual studies and systematic reviews:2

• Can chance explain the apparent subgroup effect?

The examination of multiple subgroups, along with a P value = 0.39 raises the question of a chance effect.

• Is the effect consistent across studies?

There are no randomized trials with which to compare.

• Was the subgroup hypothesis one of a small number of hypotheses developed a priori with direction specified?


• Is there strong preexisting biological support?

The result is not biologically implausible.

Thus, the reader should approach the conclusions of the authors with caution in making their therapeutic decisions.


  1. Martín-Peña A, Aguilar-Guisado M, Espigado I, Cisneros JM. Antifungal combination therapy for invasive aspergillosis. Clin Infect Dis. 2014;59:1437-1445.
  2. Sun X, Ioannidis JP, Agoritsas T, et al. How to use a subgroup analysis: Users’ guide to the medical literature. JAMA. 2014;311:405-411.