Editorial: Skin hemangiomas and occult dysraphism

(OSD) are well known to most neurosurgical practitioners: the patch of hair, the subcutaneous lipoma, the dermal sinus tract, the lumbosacral dimple, and more exotic signs, such as the caudal appendage and the so-called “cigarette burn” lesion. These stigmata are a matter of neurosurgical lore, but they have received very little rigorous analysis as tests indicative of the presence of underlying surgical disease. In the terminology of evidence-based medicine, what the clinician who must decide about further investigations wishes to know is the positive predictive value (PPV) of these cutaneous signs. The PPV is the probability of disease in the patient in whom a test is positive, in this context, in the patient with a certain physical finding. Clinical experience suggests that the PPV of the classic lozengeshaped lumbar hairy patch is very high. My impression is that nearly all such patients harbor underlying split cord malformations. The PPV of the flat, lumbosacral, capillary hemangioma (“salmon macule” or “strawberry nevus”) is much more problematic. The treacherous aspect of the concept of PPV is that it is not an intrinsic property of a test. Sensitivity and specificity are intrinsic properties, but, unless the specificity is perfect, the PPV of a test is dependent on the population of cases to which it is applied. If the prevalence of the target condition is high in the study population, the PPV will be high; if the prevalence is low enough, the false positives will swamp the true positives, and the PPV will be low. In discussions of lumbosacral capillary hemangiomas, readers must pay close attention to the derivations of the study groups. In this issue Tubbs, et al., describe 21 patients with OSD in whom the only cutaneous sign was an isolated capillary hemangioma. These 21 cases were culled retrospectively from the pooled personal series of three pediatric neurosurgeons treating a total of 120 cases. Symptoms, neurological signs, and reasons for referral are not presented. Logically, one can conclude from this report that there exist patients with isolated capillary hemangiomas in whom OSD is present. Determination of the PPV of isolated capillary hemangiomas in a defined population was not the object of this study. In contrast, we recently reported the yield of positive ultrasonographic findings in a small series of infants who were referred for pediatric neurosurgical consultation because of lumbosacral capillary hemangiomas (or strawberry nevi, as we termed the lesion in question).1 We attempted to estimate the PPV of the isolated strawberry nevus in a reasonably well defined, clinically relevant study population. Among 14 infants with isolated strawberry nevi, there were no positive spinal ultrasonographic examinations, and in an additional infant the magnetic resonance (MR) imaging study was negative. We estimated the PPV of the isolated strawberry nevus in terms of ultrasonographic yield to be 0 with a 95% confidence interval ranging to 0.2. We concluded that ultrasonographic screening of infants with isolated strawberry nevi is probably not worthwhile, and we asserted that the strength of the association between strawberry nevi and OSD remains unquantified. Reconciling our observations with the observations of Tubbs, et al., is not difficult. Because the sensitivity of spinal ultrasonography is not perfect,2 there may have been infants with OSD in our study group who escaped detection. Neonatal strawberry nevi at other locations, such as at the glabella, on the eyelid, and in the subocciput, fade variably with time. Many of the lumbosacral strawberry nevi in our study group likely faded later in childhood as well. Perhaps inconsequential lesions tend to fade and significant lesions persist, in which case the PPV of an isolated nevus in an older population would be greater than in a younger population. We do not know. Tubbs, et al., make an unqualified recommendation that lumbosacral hemangiomas be evaluated with MR imaging. This recommendation reflects expert opinion and common practice, but its evidentiary basis is very slim. The cost of such a recommendation is unknown but, because of the large number of infants with lumbosacral hemangiomas, probably quite high in terms of dollars and sedation-related misadventures. The benefit is unknown as well, in view of the undocumented natural history of