This study, in a large cohort of PKU patients, shows that the 48-hour BH4 loading test has a high PPV for true BH4 responsiveness. Furthermore, it shows that patients with genotypes including at least one mutation known to be associated with long-term BH4 responsiveness have a high rate of true response. This is the first study to investigate the PPV for BH4 responsiveness of the 48-hour BH4 loading test and the additional value of genotype. Establishing a reliable pre-selection method is of high importance, since it could save time and money, and will prevent false expectations. The treatment of BH4 responsive PKU patients has been markedly changed since the therapy was introduced. Expansion of the Phe-restricted diet or even complete liberation can reduce a major burden. Although quality of life results obtained by standardized questionnaires (not suitable for PKU) do not show a significant improvement, patients do report an increased quality of life [27, 28].
Recently, we amongst others suggested that 24 hours is not long enough to detect patients with a delayed response to BH4 [16, 31–33]. Indeed in this study we confirm that patients with true BH4 responsiveness can be missed in a 24 hour test. Other studies suggest extending the BH4 loading test to beyond 48 hours [9, 28, 34]. We detected some patients who tend to show a response at T = 48. However, we did not compare our results to a longer lasting test, such as the 7–28 day BH4 challenge. As our data indicate that patients who show a first response at T = 48 are less likely to show true BH4 responsiveness than patients who respond within 24 hours, prolongation of the BH4 loading test will possibly increase the number of patients with true BH4 responsiveness, but will also decrease the PPV and specificity. Also, influence of external factors increases with the length of the test. Moreover, patients with only one moment of response during the 48-hour BH4 loading test can show true BH4 responsiveness. Another strategy is published by Singh and Quirk. They chose to see whether it is possible to liberalize the diet in patients with just 15% decrease of Phe after one month of BH4 treatment . Unfortunately however, this study does not show whether such low a cut off results in the detection of extra BH4 responsive patients.
As expected blood Phe concentrations at the diagnosis, baseline Phe concentrations and Phe concentrations prior to the BH4 loading test were significantly lower in patients with true-positive BH4 responsiveness than patients without potential BH4 responsiveness and false-positives. Nevertheless, these biochemical parameters can give an indication but show too much overlap for practical use in predicting BH4 responsiveness. Another notable finding was that a majority of false-positive patients are found in the younger age range. At present, we do not have a clear explanation for this finding. Possibly, younger patients have more variability in their blood Phe concentrations.
Since a few years it is recognized that a genotype can be helpful in predicting BH4 responsiveness and to do so a full genotype is required [14, 15, 36]. Our results underline this assumption. Patients who have a functionally hemizygous genotype are frequently true BH4 responders. The three patients who were not potential BH4 responders all reached a 20-30% Phe concentration decrease in the 48-hour BH4 loading test and can potentially also benefit from the therapy. The patients with two mutations associated with true BH4 responsiveness were all true BH4 responders. This concerns only three patients as patients with this genotype combination probably have mild hyperphenylalaninemia very often and do not always require treatment, which was an inclusion criterion. Furthermore, the results in this study support the suggestion in the literature that patients with two putative null mutations are not candidates for BH4 treatment [36, 37]. In a recent study by Sterl et al. the proportions of predicted BH4 responsiveness are fairly similar to our results .
An important remark is that a classified genotype is needed to predict BH4 responsiveness. As more than 400 missense mutations are described and 75 to 90 percent of the patients are compound heterozygous [8, 13, 27], this results in a large number of as yet unclassified genotypes. Additionally, some mutations are associated with an inconsistent BH4 response, for instance p.R158Q [8, 13], p.I65T [21, 26, 39, 40] and p.R261Q [8, 13, 14, 27]. In our patients, where the p.R158Q mutation was homozygous or combined with a putative null mutation, it showed no BH4 responsiveness. The p.I65T mutation combined with a putative null mutation showed potential response in two out of two patients. However, unfortunately, one patient did not perform the BH4 extension phase (60% Phe concentration decrease) and the other was a false-positive. The p.R261Q mutation again showed variable results, with a response just under the threshold of 30% remarkably often. Patients who are functionally hemizygous for p.R261Q seldom show true BH4 responsiveness, but one of our patients and one patient described by Hennermann et al. , both p.R261Q/L.P281L, show (modest) long-term BH4 responsiveness.
Since this study reports patient care results, patients without potential BH4 responsiveness did not participate in the BH4 extension phase. This limits the conclusion regarding its diagnostic value. Other results, however (number of false-positives in patients with borderline response), support the presumption that the number of false-negatives will be limited. A possible bias in the 48-hour BH4 loading test could be the supplementation of Phe to increase the baseline Phe concentration and the different methods to reach this increase. The influence of the different metabolic kinetics is unclear [41, 42]. With respect to the true BH4 responsiveness criteria, the extra 4 g of natural protein (180 mg Phe) tolerance increase criterion can be considered relatively modest. This age independent criterion was agreed within the Dutch group of metabolic physicians. If the criterion were doubled to 8 g of natural protein, four patients would not have met this tolerance requirement. Two out of these four patients also showed a 30% or more decrease in blood Phe concentration. Of the other two patients (both children), genotype showed heterozygosity for p.R261Q/p.P281L in one and c.1315+1G>A/p.L348V in the other. Furthermore, we anticipated that the BH4 extension phase would take less than six months. However, the actual duration exceeded this six months time period in 41 percent of the patients, with three patients taking even longer than one year. There are several explanations for this longer duration of which the exploration of the possible speed of Phe tolerance increase is an important one, as is the motivation of the patients during the last stages (especially blood sampling frequency).