Letters to the Editors

Sirs, The recent paper by Clark et al. (1998) highlighted many issues regarding the interpretation of short synacthen tests (SST). In 1995 as part of an extensive pre-launch evaluation of the cortisol method on the Bayer/Chiron ACS180 (ACS) we compared responses to SST with two established methods (Abbot TDx, nˆ 45 and Farmos, nˆ 46). Like Clark et al. (1998) we noted the non-gaussian nature of the data and used non-parametric regression analysis (Thiels) and found the slope and intercept were signi®cantly different in samples 30 minutes post synacthen (SST30) [ACSˆ 0 ́79*TDX‡ 110; ACSˆ 0 ́79 *Farmos‡ 176] compared to basal (SST0) [ACSˆ 0 ́86*TDX ‡51; ACSˆ 1 ́08*Farmos‡ 14] (Price & Ross, 1996). 105 samples taken from patients on hydrocortisone treatment for adrenal insuf®ciency showed no dose dependant bias (ACS180ˆ 0 ́96*TDX‡ 18). Our theory has been that in response to synacthen the adrenal produces a short lived metabolite which cross reacts in the TDX and Farmos assays but not the ACS180 assay. Such an explanation could be applied to the data found by Clark et al. (1998). We thought the lower values post synacthen would affect the interpretation of the SST so, using a peak cortisol response of 550 nmol/l or greater as adequate (Lindholm et al., 1978; May et al., 1985; Stewart et al., 1988; Axelrod et al., 1993; Howlett et al., 1994; Jackson et al., 1994; Tietz et al., 1995), we compared the results from the ACS with the other methods and found only 1 out of 95 patients to give a discrepant response. This was in a male aged 54 with chronic renal failure. The other 3 patients with chronic renal failure for whom we had data did not give discrepant responses. Therefore, the lower results produced by the ACS has a minimal effect on the interpretation of the SST. Previous investigators may have found a change in incremental response in SST with basal cortisol because they did not account for a difference in response between males and females noted by Clark et al. (1998). We re-analysed our data (Price & Ross, 1996) for both the Farmos and TDX methods and found no signi®cant change in the incremental response of cortisol against the basal concentration for males but found a signi®cant (P < 0 ́01) decrease in incremental response against basal concentration for females. Wood (1998) highlighted the need for rationalization of SST. However, why stop at SST? The problem for laboratories and manufacturers of cortisol methods, is the multitude of methods that are used by clinicians to assess adrenal reserve. The problem is further exacerbated if the de®nition of an adequate response for one method is linked to the demonstration of an adequate response for another, for example SST and the insulin stress test (IST) (Lindholm et al., 1978; Stewart et al., 1988). This raises the question of whether it is valid to establish the cut off for the SST without reference to results from IST in the same patient using the same cortisol method. A multi-centre evaluation of tests used for adrenal reserve is required. For example, a clinician performing an IST would also do a SST and a glucagon test. Suf®cient serum would be taken at agreed time points and stored in multiple aliquots at ÿ808C allowing responses to be established for a variety of cortisol methods for a variety of adrenal insuf®ciency tests all linked to the gold standard of IST. Suf®cient serum should be taken that many surplus aliquots are available and at the end of the trial these should all be transferred and stored at one site. Manufacturers who then produce a new cortisol method would have a reference laboratory where they can evaluate and produce their own response cut off for the variety of methods. Such a multi-centre evaluation is needed for the bene®t of patients who have to undergo these tests, for clinicians and for laboratories who have to try and interpret them. Alun Price*, T. A. Gray* and A. P. Weetman2 Departments of *Clinical Chemistry and 2Medicine, Northern General Hospital, Shef®eld, U.K.