Defining clinical failure for salvage studies

In patients with invasive aspergillosis (IA), there are numerous clinical settings where stable disease or progression of findings or deterioration of the patient's condition does not indicate failure, and subsequent response to a 'salvage' antifungal is not necessarily attributable to this drug. Many patients, in whom pulmonary aspergillosis emerges during profound neutropenia, show enlargement of their lesions on computer tomography (CT) scans, eventually accompanied by clinical deterioration, during hematopoietic recovery. This may in fact represent the recruitment of neutrophils and monocytes to the pulmonary 'battlefield', resulting in a favorable clinical outcome also without changing antifungal treatment. Infarcted tissue may contain vital filamentous fungi, because it is poorly penetrated by the antifungal, not indicating a lack of efficacy of this drug against the respective fungal pathogen. In patients treated with an echinocandin, serum galactomannan levels may increase despite successful treatment. Piperacillin-tazobactam or other semi-synthetic beta-lactam antibiotics may cause false positive serum Aspergillus galactomannan levels. Patients primarily treated with a lipid formulation of AmB (LF-AmB), who are switched to a 'salvage' antifungal, will unequivocally receive a combination therapy due to the persistence of high LF-AmB concentrations in tissue. Criteria to define 'clinical refractoriness', 'resistance', 'non-response' or 'failure' respectively should be re-defined. One option to establish a more valid definition would be to use a composite score including clinical as well as radiological and microbiological or mycological criteria. The latter may include non-culture based methods such as serum galactomannan. Assessment should not be made earlier than after seven days of full-dose systemic antifungal treatment. However, in individual cases, e.g. a patient with hematopoietic recovery, who shows an increasing volume of pulmonary aspergillosis and clinical deterioration, it may be recommended to refrain from switching the patient to another regimen and continue the current antifungal treatment for another seven days before failure is stated. Clinical studies on second-line antifungal treatment for IA should be randomized and blinded, patients should be evaluated separately with respect to their reason for 'failure' of primary antifungal treatment, and stratified according to their previous antifungal treatment. Ideally, the first-line regimen would be standardized. Host criteria such as neutropenia or graft-versus-host disease (GVHD) should be clearly defined and documented with respect to their course over time, and patients should be stratified according to these criteria. A three-arm study (continuation of primary antifungal vs. combination of primary antifungal with a 'salvage' drug vs. the 'salvage' drug alone) would be ideal.