In conducting this evidence-based review, the panel’s goal was to provide practical guidance to busy clinicians, using the RAND/UCLA Appropriateness Method to help synthesize evidence and experience. We found that the evidence base for much IPF treatment is suboptimal, with the clearest evidence suggesting 1) the need for additional randomized trials, 2) early referral of eligible patients for lung transplant evaluation, and 3) no role for corticosteroids alone in cases of confirmed IPF.
The utility of the RAND/UCLA method is most easily identifiable in areas where the definitive randomized, double-blind clinical trial has not been conducted, but the overall weight of the evidence is compelling. For example, no randomized trial has ever been conducted comparing corticosteroids alone to placebo in patients with confirmed IPF (defined using recent criteria). However, the data that are available show no objective clinical benefit (and only transient subjective benefit) from the use of corticosteroids in clearly-defined IPF [9
]. Hence in reviewing the body of evidence, the panel had consensus in concluding that for cases of confirmed IPF, corticosteroids alone have no role as the sole, initial therapy. Three other important themes emerged without disagreement.
First, if a prospective clinical trial is available, referral for participation was always considered appropriate. Until there is a definitive treatment for IPF, a randomized trial may be one of the best ways for a patient to receive potentially beneficial treatments, while advancing the state of the science for future patients with IPF.
Second, the appropriateness of referral for allogenic lung transplantation depended upon the patient’s age: inappropriate in those aged 65 or older, and appropriate for younger individuals. Lung transplant is not currently an option for most patients over 65 in the United States, though some centers will consider individual patients on a case-by-case basis.
Third, although preliminary data appear promising, the panel rated the use of pirfenidone as “uncertain” [12
]. This drug is not currently routinely available to US physicians, and at this time it is not possible to make a recommendation regarding the appropriateness for clinical practice. At the time of the literature review data were also lacking for other agents being investigated for efficacy in IPF, including bosentan, thalidomide, etanercept, and imatinib.
Two themes emerged with intermediate levels of disagreement: for the use of azathioprine and corticosteroids, and interferon gamma-1b.
Initiating therapy with a trial of azathioprine and corticosteroids was rated appropriate by the panel for patients under age 65, with mild-to-moderate impairment, who had never been treated or had failed steroid therapy alone. Prospective clinical trial data comparing azathioprine to placebo were not available. One randomized trial comparing azathioprine and corticosteroids to corticosteroids alone in 27 newly-diagnosed patients with IPF found no significant differences in the primary analyses for lung function, oxygenation, or survival. The secondary analyses adjusted for age found a survival advantage for the group that received azathioprine [14
]. The panelists favored azathioprine over cyclophosphamide because of the potential differences in side effects of the two drugs.
Use of interferon gamma-1b as first line therapy was deemed uncertain for those under 65 and inappropriate in patients over 65 with severe impairment. The panel rated the use of interferon gamma-1b as appropriate – not as first line therapy – but only after initial management had failed in patients with mild-to-moderate impairment (FVC > 55% and DLco
> 35%). A recent, large randomized trial of interferon has failed to demonstrate significant efficacy compared to placebo [15
]. Efficacy was based on the combined primary outcome of death or disease progression (defined as an FVC decrease of 10%, A-a gradient increase of 5mm Hg, or death) which was not significantly different for the group randomized to IFN-γ1b relative to placebo control (OR 0.80 (0.52, 1.24)). However, subgroup analyses, which are generally considered at best hypothesis generating, showed a possible survival benefit for patients with mild-to-moderate impairment at time of randomization.
The RAND/UCLA Appropriateness Method
The validity and usefulness of guidelines depend on the method used. We used the RAND/UCLA Appropriateness Method, a well-studied approach that blends evidence from the literature with an expert consensus process [7
]. No panel can deduce the results of a definitive randomized controlled trial before it has been conducted. It can, however, assess what is known today in as formal and unbiased fashion as possible. The panel received an extensive systematic review of the literature (see for the studies reviewed in detail). The systematic review included data on the potential benefits and risks of treatment options for IPF. At the time of the panel review, results from the IFN gamma-1b randomized trial were available as the results and subgroup/exploratory analyses had been presented by some of the panel members at scientific meetings. Thus, the panel had an opportunity to consider these most recent data in detail.
Studies included in systematic literature review provided to panelists
Consistent with the RAND/UCLA Appropriateness Method, the panel had a diverse US composition. The panel included representatives from academic as well as community-based practice. The panelists represented all major geographic regions of the US, and clinical practice in the US tends to be similar to that in Europe for this disease. The panel reflected multiple viewpoints and this mitigates potential geographic and practice-type bias.
The development of the guidelines used a quantitative process. Each panelist rated 324 clinical scenarios on a 9-point risk-benefit scale from least appropriate (1) to most appropriate (9). These scenarios reflect real-world combinations of clinical factors. The final ratings were calculated as the median scores among the nine panelists, not necessarily reflecting each panelist’s opinion. When statistical disagreement occurred (i.e., at least 2 panelists rated a scenario appropriate and at least 2 panelists rated that same scenario inappropriate), that scenario was given a rating of uncertain.