To our knowledge, this study is the first to evaluate endothelial function using PAT in NF1 patients. Contrary to expectations, these results suggest that the endothelial function of young adults with NF1 does not different from that of similar subjects without NF1.
Exacerbated vascular injury responses and chronic inflammation have been observed in murine models of NF1, and chronic inflammation in patients with NF1, suggesting that these factors could lead to ED . Neither that study, nor studies on neurofibromin and endothelium in mice  and humans  tested endothelial function, although they did show that neurofibromin was expressed in endothelial cells  and that neurofibromin was important in cardiac embryogenesis in mouse models .
A death certificate analysis  reported a lower proportionate mortality ratio (PMR) for diseases of the circulatory system in individuals with than without NF1, especially for ischemic heart disease, which is closely related to atherosclerosis and endothelial dysfunction [15, 16]. Moreover, in a similar study , although the PMR for vascular diseases in general was found to be higher in NF1 patients, the PMR for diseases of the arteries and arterioles, including atherosclerosis, was lower, a finding consistent with ours.
Our method of assessing endothelial function has several limitations. However, another method, the brachial artery reactivity test (BART), while being the most validated, requires extensive training and standardization. This has led to the development of alternative methods of evaluating endothelial function , including the PAT used here. We chose this method because it is simple, non-operator-dependent and commercially available, and because our research group has used it in patients with other diseases and has found its results reproducible .
Moreover, PAT and BART do not measure the same aspects of vascular biology. PAT measures the vasodilator response of the digital microvasculature to ischemia, whereas BART measures the response of the conduit arteries. The vascular bed and vessel size likely determine sensitivity to early damage from specific cardiovascular risk factors [19, 20]. Therefore, the stage of disease progression may affect the ability of different methods to detect “abnormal” endothelial function. Nevertheless, both the PAT and BART methods correlate well with risk factors for CAD [15, 19] and with coronary artery endothelial function [12, 21] and both are able to predict future vascular events [16, 22], although they correlate modestly, if at all, with each other . Therefore, each method complements the other, providing additional information on endothelial function. BART likely provides more important information on individuals with existing atherosclerosis, whereas PAT may be an earlier indicator of CAD risk. Thus, PAT seemed more appropriate for profiling our study population.
Another concern about the PAT method was that the cutoff value was based on findings of a receiver operating characteristics (ROC) curve, which found that a cutoff value of 1.35 to diagnose patients with ED had a sensitivity of 80% and a specificity of 85% . This cutoff value was based on patients referred for diagnostic percutaneous coronary angiography, a high-risk population with current or developing vascular disease. Thus, using the same cutoff value on a population of young, asymptomatic adults without vascular disease risk factors, as in our study, is questionable because it could change the predictive value of the test, especially if used for screening. We therefore used a combination of this cutoff value and LnRHI, which allowed for a comparison of mean values. Nevertheless, we found that the between group difference in mean LnRHI values was not statistically significant.
Only a few studies have used EndoPAT to non-invasively assess endothelial function in young subjects, including children, making it difficult to determine the expected prevalence of ED in the control group. One study tested the feasibility and reproducibility of EndoPAT in adolescents but did not indicate the percentage of subjects that presented with ED . Similarly, a study that assessed ED in diabetic children  did not indicate how many of them had ED, as defined by the cutoff values. Since both studies compared mean and standard deviations of RHI values between groups, it was difficult to determine how many subjects in each group were below the cutoff value. Moreover, when establishing values in healthy control subjects, it should be remembered that the results of non-invasive endothelial function tests vary significantly in an individual on different days; however, it is not clear if this is due to physiologic oscillations in endothelial function or to the imprecision of the methods .
Several characteristics differed significantly between our NF1 and control groups, including height, body mass index of women and resting heart rate, but, after adjustment to account for these differences in the covariance analysis, none was found to affect RHI (data not shown).
One of the relative limitations of this study is its sample size; the observed difference in the prevalence of ED (10.5%) was lower than the difference expected (30%) when calculating sample size, thus reducing its post hoc power (30%). Although this sample was not powered to exclude small differences between groups, our findings suggest that ED is not clinically relevant in patients with NF1. This lack of difference in endothelial function suggests the need for further studies evaluating additional markers of ED, markers that are more sensitive and specific than PAT. In addition, older patients with NF1 should be included. These changes may result in a more precise determination of the function of neurofibromin in haploinsufficient endothelium.
Finally, understanding the natural history of NF1 vasculopathy requires prospective evaluations of the endothelial response of NF1 patients to risk factors of vascular disease. Our clinical observations suggested that atherosclerotic vascular disease risk factors, such as obesity, type 2 diabetes and hypertension, have lower prevalence in NF1 patients than in healthy controls, although this finding has not been confirmed. Thus, NF1 patients who die at a younger age due to vascular disease may do so from pathological mechanisms that may be different from those in the general population. Vascular disease in patients with NF1 may be caused by other factors, such as arterial fibromuscular dysplasia, rather than atherosclerotic disease . Furthermore, both death certificate studies showed that patients with NF1 had a lower PMR for atherosclerosis-related conditions, including ischemic heart disease  and diseases of the arteries and arterioles , and a much lower PMR for atherosclerosis risk factors, such as diabetes [3, 4] and hypertension .
These observations and our results are in agreement, that ED is not more prevalent in patients with NF1 than in the general population. Prospective assessment of vascular disease and endothelial function in patients with NF1 may provide greater understanding of this subject.