Health status decline in α-1 antitrypsin deficiency: a feasible outcome for disease modifying therapies?

Sample size calculations for a study with SGRQ as the primary outcome together with potential physiological outcomes were based upon a two group parallel comparison, using the formula n = 1 + 2C (S/D)2 where D was the smallest difference to be detected (a 25% change in SGRQ decline was used as an arbitrary effect size for illustrative purposes) and s represented the standard deviation of the observations. C = 10.51 to provide a 90% power of detecting a change in SGRQ at the 5% level of significance. As the deterioration in HS was determined by slope analysis over at least 4 annual assessments, power would be relevant for a 3 year study.

Data was collected from 454 patients with AATD (PiZZ) with median follow up of 7 years (IQR 5–10). Demographics of patients with and without airflow obstruction (FEV₁ /FVC ratio greater or less than 0.7) [18] are shown in Table 1. The majority with airflow obstruction were male (64%) and 89% were index cases diagnosed due to respiratory symptoms. The majority were ex smokers (81%):- no current smokers were included in the analysis as there were too few to analyse meaningfully. Patients without COPD were mainly female (61%), non index cases, never smokers and younger (Table 1). There was a wide range of pulmonary function from normal to very severe COPD (GOLD stage 4) also reflected in the gas transfer results and SGRQ scores for the groups with and without COPD.

The baseline SGRQ score for the whole group showed a statistically significant (p < 0.0001) correlation with baseline FEV₁ although the r² value of 0.34 indicated that only 34% of the variability in the group was explained by FEV₁ alone (Fig. 1). The correlations for those with and without COPD were also statistically significant (p < 0.0001) although the variance for each group was less (r² = 0.20 and 0.15 respectively). Similar statistically significant correlations were found between SGRQ total score and gas transfer for the whole group (r² = 0.105; p < 0.0001) and those with COPD (r² = 0.015; p = 0.008) but not those without COPD (r² = 0.002; p = 0.35).

There was a wide range of FEV₁ and Kco decline (% predicted/year). Generally the SGRQ scores increased yearly for the groups with and without COPD, although again a wide range was seen (Table 1). The symptom and activity score was stable for the non COPD group even though the median total score showed gradual deterioration (+ 0.21 units /year). The group with COPD showed a worsening of symptoms and activity with time and an overall median increase in total score of + 0.66 units/year although only the activity and total score deteriorated more rapidly than for those without COPD (p = 0.001 and p = 0.025 respectively). Figure 2 shows the annual change in SGRQ and its domains in patients with and without COPD, divided according to no or rapid decline in FEV₁ [19]. Summary data and demographics of patients in each of these groups is provided in Additional file 1: Tables S1 and S2. In addition the FEV₁ and Kco decline is summarised in Additional file 1: Tables S3a, b and c for comparison with previous literature that did not account for age sex and height.

There was no difference in the rate of gas transfer decline between those with and without COPD (p = 0.125). There was no relationship between SGRQ decline and gas transfer decline (Additional file 1: Tables S4 and S5). The decline in FEV₁ was statistically related to the increase in SGRQ total score for the whole group (Fig. 3; p < 0.0001), although the relationship (r² 0.04) indicated only 4% of the decline in HS was explained by the decline in FEV₁ alone. The decline in HS was greater in COPD patient with rapid decline in FEV₁ (%predicted) compared to those with no decline although only for the total score and activity domain (p < 0.05) as shown in supplementary e-Table 2.

The decline in SGRQ total score did not differ between GOLD stages (Fig. 4).

Forty eight % of patients without COPD and 2% of those with COPD were not receiving long acting inhaled therapy. The proportion receiving LABA, LAMA or ICS alone were similar for those without and those with COPD (11% vs 19%: 9% vs 9%: and 4% vs 7% respectively). However, only 9% of those without COPD were on triple therapy (LAMA/LABA/ICS) compared to 61% of those with COPD. The high proportion of those without spirometric COPD at the time of referral who were on regular inhaled therapy largely reflects clinical practice at the time of referral and symptoms related to discordance beteen spirometry and gas transfer as described previously [21]. There was no relationship between the decline in HS and the nature of regular inhaled therapy assessed as categorical data.

Table 2 summarises the numbers needed to confirm a 25% difference in decline in SGRQ total score or FEV₁ for the group with COPD, and for those with rapid decline in FEV_1, assuming 90% power and with a type 1 error set at 5%, The numbers required to detect a 25% difference in Kco decline, in those with COPD and rapid decline is also shown.

Patients were divided into those with and without airflow obstruction and all SGRQ domains and total score showed a wide range of values though significantly worse for those with COPD. However, even in those without COPD the total score and especially the symptom domain score was higher than expected in a healthy population [22]. This likely reflects the proportion of patients in our study who were “index” cases and hence diagnosed after presenting with symptoms. Whereas these patients did not have COPD by spirometric criteria we have shown previously that in AATD the gas transfer can be reduced even in subjects with normal spirometry and likely leads to referral and assessment [21]. This concept is consistent with the lower scores in the non index Swedish birth cohort who have been followed for up to 40 years where HS is maintained and only a small number of current smokers show HS or gas transfer impairment [23].

Changes in HS were greater in those with established COPD, especially the activity score which increased almost + 1.2 units/year, whilst total score change was lower (+ 0.66/year) reflecting the lesser changes in symptom and impact domains. The relationship to GOLD stage indicated those with mild and moderate COPD had the greatest median deterioration in total SGRQ score although there was no statistical difference across groups. Our previous studies had shown little change in average SGRQ scores over 3 years [24] which was at variance with a Dutch AATD study which showed a major decline in total score (+ 6.5 units) over 30 months [25]. The difference likely reflects a combination of the centralisation of patient care and the degree of monitoring carried out at our centre as well as potential improvements in patient care over the intervening 14 years. More recently a study from Germany [26] also documented a slow decline in SGRQ in AATD (+ 1.2 units for total score) which weakly reflected annual exacerbation rates [26]. However, that data was self-reported and included unquantified patients with the SZ phenotype and a significant proportion on augmentation therapy, which may have influenced the results.

The SGRQ domains and total score showed little change in patients with no airflow obstruction. However there was a relationship to the FEV₁ decline (% predicted): those with rapid decline showed a greater median increase in all domains and total score although the range was wide and only significantly different for the symptom domain. There were differences in the activity domain and total scores when the COPD patients were divided into those with no or rapid decline in FEV₁. However, when divided into those with rapid or no decline in gas transfer, although the activity domains rose by median values of + 0.18 and + 1.17 units/year respectively, neither achieved statistical significance (Additional file 1: Tables S4 and S5) suggesting the rate of FEV₁ decline has the more noticeable effect on HS. Although there was an overall (significant) correlation between FEV₁ decline and SGRQ deterioration, the relationship was weak (Fig. 3).

These data have important implications for clinical trials of disease modifying therapies in AATD and non deficient COPD. Age itself (without COPD) has an effect on HS [22] although clearly from a lower baseline than in COPD. AATD is considered a more rapidly progressive type of COPD but even those with established COPD stratified for rapid FEV₁ decline had an overall median annual increase in total SGRQ score that was only + 0.5 points higher than those with no age corrected FEV₁ decline. Indeed even when stratified into moderate and severe COPD (typical of patients currently included in clinical trials) the differences remained small and statistically insignificant.

Our data was generated from a specialised research programme with annual follow up, which conceivably has a similar placebo effect on HS to that seen in clinical trials [27] and may reflect changes in patient expectations as well as close monitoring and better adherence to background therapy [28]. Despite similar close monitoring, many interventional trials in COPD demonstrate a greater treatment effect on HS parameters (≥ MCID of 4 points for SGRQ total score) which may reflect the nature of the trials aimed at both short term improvement in lung function and modification of exacerbations (e.g. [29, 30]). However, in trials of augmentation therapy in AATD to date no signal has been detected either statistically or close to the MCID [13‐15] leading to scepticism about efficacy despite significant effects on emphysema progression [31]. This failure may reflect both the size of the studies in this rare condition and the lack of any immediate improvement in lung function (and hence a noticeable clinical signal) for disease modifying treatments like augmentation.

We observed a slow deterioration in SGRQ scores in patients not receiving augmentation therapy, indicating that detection of slowing of HS decline in clinical trials aimed at disease modification is likely to require very large numbers for adequate power. If augmentation stopped all HS decline it would still require at least a 4 year study in rapid FEV₁ decliners to achieve the MCID for total SGRQ score. Our data shows that even those with stable FEV₁ have an annual deterioration in HS which is half that of the rapid FEV₁ decliners indicating that a study to find a median 4 point effect of treatment that stopped the physiological related decline would take about 8 years.

In the current study we calculated the number needed to confirm a 25% reduction in SGRQ total score decline which, even in pre-identified rapid decliners with COPD (5039 / arm) is prohibitive for a “rare” disease, especially when most identified patients are already on widely accepted AAT augmentation therapy. To detect an SGRQ decline reduced to that seen in COPD patients with a stable age related decline (50% reduction) would still require 1260/arm if treatment were totally effective. Using slope analysis allows projections of the longer term benefit on HS decline, but would still require at least 3 years for decline to be documented and would still not reduce patients required in such a trial to a feasible number.

Slope analysis of health status was preferred to endpoint analysis to overcome any potential regression to the mean resulting in an apparent improvement in scores overtime [32] and at the same time allow more robust longitudinal projection of the changes to determine time to MCID.

Since SGRQ appeared prohibitive as a primary or secondary outcome we also conducted power calculations for a prospective study using both FEV₁ and Kco as the outcome. Again the numbers appear prohibitive for unselected patients with AATD and COPD. However, the numbers become more feasible if patients were chosen by known rate of rapid physiological decline (Table 2) i.e. enriching the study population for the outcome being assessed. This approach could be considered for both a pragmatic clinical trial and, in the interim, as a criterion for assessing treatment response [19]. Interestingly the power calculations form a rank order consistent with previously published data on the sensitivity of these 3 parameters to change [33].

The current study has a potential weakness. It represents a cohort of patients referred to a specialist centre for annual long term follow up as a research programme. Thus it will not represent “usual” clinical care and a degree of selection bias cannot be discounted. However, this is also a strength as the patients and their management likely represent those who would be suitable for, and included in, clinical trials. The results therefore represent what impact a focussed clinical service has on HS and its deterioration and are therefore relevant to the design of trials of disease modifying therapies in AATD where placebo arms should achieve similar results. Equivalent data will also be important in non AATD COPD as use of disease modifying therapies expand to this patient group.

As we were studying HS decline we only included patients in whom this had been documented on at least 4 annual visits. Some patients may not have had this recorded and some may not have returned on at least 4 occasions either because they had become reassured of their physiological “normality” of a combination of distance travelled with and without increased morbidity with time. Nevertheless this selections criterion did not differ with disease severity suggesting it is generally reflective of the patient groups.

The insensitivities of both group physiological data and HS, to change [33] has been an important driving force in the development and validation of CT densitometry as the outcome of choice in emphysema trials especially in AATD. Although lung densitometry relates to both FEV1 and SGRQ cross-sectionally and remains the best predictor of Mortality [34] only the decline in FEV1 has been shown to relate to the decline in lung Density [35]. Whether density decline shows a similar relationship to SGRQ deterioration has not been assessed but such studies may provide indirect evidence of any putative, clinically apparent, benefit that would not be detected by current trial design and power.