Statistics from Altmetric.com
Recombinant human DNase is an expensive mucolytic which does not benefit all patients with cystic fibrosis. Company sponsored trials in unselected cystic fibrosis patients have documented wide variability in spirometric responses to the drug, but the data are presented in a way which prevents the clinician from assessing which patients are likely to benefit.
We therefore read with interest the editorial by Dr Innes regarding the assessment of response to DNase in cystic fibrosis.1However, whilst we agree that it is necessary to target DNase, we have reservations regarding the use of “n-of-1 trials” for this therapy. Dr Innes states that this approach has been used in Scotland and quotes a study unpublished at the time of writing in support of it. However, this study has already been heavily criticised since many patients refused to take part and others did not complete the trial periods.2 Furthermore, such studies are inherently time consuming and resource intensive.
We have adopted a different approach to ensure that DNase is prescribed in a rational fashion. Before it became available on the NHS we met with local purchasers to define selection criteria and a trial protocol. Following selection, those who have an improvement in forced expiratory volume in one second (FEV1) of ⩾10% after a trial of DNase are defined as “responders” and remain on the drug. A review at two years has shown that, whilst responders maintain their improvement, non-responders are not disadvantaged.3 Thus, using this protocol we have been able to target DNase to those patients who obtain maximum benefit. This model has now been widely accepted by purchasers for adult and paediatric cystic fibrosis services in North Wales and the Northwest of England and, as such, we have no problems in obtaining funding for this very expensive product.
We suggest that Dr Innes and his colleagues abandon their “n-of-1 trials” and adopt our protocol for the use of DNase.
author’s reply We would urge Drs Ledson and Walshaw, before suggesting we abandon the Scottish n-of-1 DNase assessment protocol, to read it!1-1 The protocol has not been “heavily criticised” as they claim since the article they quote1-2 was also written without knowledge of the results (only recently published). In our experience it is very unusual for patients to refuse to undergo our assessment process.
Where we do agree is in the need to test the DNase response in individual patients. However, we disagree on how this should be done. Ledson and Walshaw advocate an unblinded, open label, two week trial of DNase using an increase in FEV1 of >10% as the only end point. We contend that this is less than ideal because (a) double blinding and placebo control periods are needed to obviate bias, given the high expectations generated in patients and carers by new treatments for cystic fibrosis; (b) using the single end point of increased FEV1 may be less reliable than combining this with other measures including exercise capacity, oxygen saturation and symptom scores; (c) a criterion of a >10% increase in FEV1 is inherently unreliable since day-to-day variability in FEV1 is around 160 ml (95% CI) regardless of the magnitude of the FEV1,1-3 so it is easy for patients with a low FEV1 to achieve an increase of >10% by chance. Indeed, Ledson et al in the description of their own protocol1-4 quote the day-to-day variability of FEV1 in cystic fibrosis as “up to 13%”, so clearly some 10% increases will be spurious. We agree that DNase can and should be targeted to maximise benefit, but feel that this targeting should be made as objective as possible. This may be laborious for doctors, but it is not nearly as laborious for patients as consigning a non-responder to long term daily nebulised therapy on unreliable evidence.