Pulmonary function
ILD are usually characterised by a restrictive lung function, i.e. a reduction in lung volumes with preserved Tiffeneau index (or FEV
1/FVC ratio) together with a reduction in D
Lco. However, in early disease, lung volumes and D
Lco may be within the normal range [
4]. Furthermore, in sarcoidosis and in Histiocytosis X evidence of airflow obstruction is also found in more than a quarter to half the patients in some studies [
4]. Lung volumes may be relatively preserved in smokers with IPF possibly due to coexisting emphysema although the Tiffeneau index remains normal. This suggests that the using a restrictive lung function as an exclusive diagnostic biomarker for ILD is neither sensitive nor specific enough [
4].
Serial lung function testing is used to monitor the clinical course of the disease [
4]. VC and D
Lco are the most used and the simplest indicators to measure change in ILD [
4]. There is, however, little agreement about how frequently these lung function measurements must be obtained in the follow up of the different forms of ILD. This is partly because the clinical course of the different ILD shows a wide variation, from acute alveolitis due to amiodarone to chronic fibrosis by scleroderma.
Different studies have addressed the question if lung function variables can be used to predict survival. A study of our group presented the survival rates between the most common forms of ILD, with a mean survival after 5 yrs of 91.6% in SARC, 84.1% in HP, 69.7% in CTD, 35.4% in IPF, 85.5% in other IIP and 69.5% in undefined forms of lung fibrosis [
32]. A Cox regression analysis shows that an age of less than 66 year, a diagnosis of an ILD that is not an IPF, a VC of more than 63% predicted and a % macrophages of less than 63% in BALF is indicative in this model of a lower mortality risk. D
Lco was not found as an independent variable for survival.
The spinoff of the IFIGENIA trial aimed to quantify the risk for mortality in 155 IPF patients after 4 years of the date of inclusion in the IFIGENIA study [
33]. The study subjects were followed with measurements of VC, TLC and D
Lco at baseline, month 6 and month 12 after inclusion. A TLC > 62% predicted (HR 0.49; 95%CI 0.30-0.81) and a D
Lco > 43% (HR 0.37; 95%CI 0.23-0.61) at baseline were found to influence rate of survival by a Cox's regression analysis. Changes in VC or D
Lco over 6 or 12 months were not found to be independent variables of survival.
Martinez et al. analysed retrospectively data from the placebo group (168 IPF patients with mild to moderate disease) of a trial evaluating interferon gamma as treatment for IPF [
6,
34]. At 12 week intervals D
Lco and VC were measured over a median period of 76 weeks. Twenty one percent died, of which IPF (89%) was the primary cause of death [
6]. In patients with an IPF-related death, 15 (47%) deaths were categorised as acute or abrupt and 16 deaths (50%) were considered subacute. For patients who survived to week 72, the mean percentage predicted FVC decreased from 64.5% (SD 11.1%) to 61.0% (SD 14.1%). The mean percentage predicted decreased from 37.8% (SD 11.1%) to 37.0% (SD 19.9%) [
6]. For patients who died during this trial, a general trend toward a decrease in FVC and D
Lco was observed, although significant intra-patient variability occurred over time [
6]. Martinez et al. concluded that the clinical course of patients with mild to moderate IPF was characterised by minimal physiologic deterioration as measured by FVC and D
Lco over a period of 76 weeks [
6].
Flaherty et al. examined retrospectively 80 patients with IPF and 29 patients with NSIP [
35]. They found that in a multivariate Cox proportional hazards model controlling for histopathologic diagnosis (UIP versus NSIP), gender, smoking history, baseline FVC, and 6 month change in FVC, a decrease in FVC of more than 10% remained an independent risk factor for mortality (HR 2.47; 95%CI 1.29-4.73) [
35].
Various other investigators have also suggested that a decreased FVC at baseline identifies patients at subsequent risk of mortality [
36‐
40]. In addition, some suggested that a decrease in FVC of 10% or more after 1 year predicts mortality in patients with IPF [
41]. Spirometric assessment is particularly valuable as its measurement is standardised [
43] and the variability in FVC is well defined among normal subjects and patients with pulmonary disease [
42,
43]. Despite these standards, a variability in the FVC measurements of patients with IPF over time is noted [
44,
45]. As a result, a wide variety of thresholds for change in FVC is used, including changes ranging from 10 to 15% in FVC [
35,
39,
41,
46‐
48].
Flaherty et al. demonstrated that the change in D
Lco over 6 months of follow-up has limited prognostic value and this measurement was not found to add independent predictive value for mortality in IPF [
35]. Although standards are presented for its measurement [
49,
50], the D
Lco varies to an even greater extent than FVC and clinically significant changes are believed to be more than 20% [
35,
41,
46,
47,
51]. As such, a survival advantage was noted by one group in patients with an improved or unchanged D
Lco compared with those experiencing a decrease of 20% or more after 1 year of therapy [
35].
The question remains whether FVC and D
Lco are useful as surrogate markers for survival in trials searching for therapeutic strategy in IPF? This question is critical since change from baseline to a specified time point of these physiologic variables has been used either or both as primary endpoint in the most recent therapeutic trials in IPF [
52].
The INSPIRE trial, the largest trial ever, included 1373 IPF patients, and concluded that no difference is seen in survival between those treated with interferon gamma and those with placebo [
53]. Consistent with the findings as described above, the INSPIRE investigators noted negligible changes in mean values of FVC and D
Lco during 77 weeks treatment with either interferon gamma or placebo [
53].
With the available evidence, it is difficult to conclude that a decrease or a change over time in FVC or D
Lco are valid biomarkers for survival, because the results of above mentioned trials are confusing. Is survival itself the only valid endpoint to be used in therapeutic trials? IPF is a rare disease, and if investigators decide only to use survival as primary endpoint, trials will be difficult to set up and manage to a successful conclusion. Taking the INSPIRE study as an example, the investigators calculated at the onset of the trial that to have 90% statistical power to detect a treatment effect equivalent to a 50% reduction in 3-years, about 600 patients were needed to achieve the targeted number of deaths within the planned duration of the study (77 weeks). Two interim analyses of this trial increased the number to 1200 patients. The economic and logistic efforts (worldwide participation of 82 centres) to bring this to a successful conclusion are huge. Efforts not every investigator or pharmaceutical company can afford [
54].