Endoscopic third ventriculostomy with choroid plexus cauterization (ETV/CPC) offers an alternative to shunt treatment for infantile hydrocephalus. Diagnosing treatment failure is dependent on infantile hydrocephalus metrics, including head circumference, fontanel quality, and ventricle size. However, it is not clear to what degree these metrics should be expected to change after ETV/CPC. Using these clinical metrics, the authors present and analyze the decision making in cases of ETV/CPC failure.
Infantile hydrocephalus metrics, including bulging fontanel, head circumference z-score, and frontal and occipital horn ratio (FOHR), were compared between ETV/CPC failures and successes. Treatment outcome predictive values of metrics individually and in combination were calculated.
Forty-four patients (57% males, median age 1.2 months) underwent ETV/CPC for hydrocephalus; of these patients, 25 (57%) experienced failure at a median time of 51 days postoperatively. Patients experiencing failure were younger than those experiencing successful treatment (0.8 vs 3.9 months, p = 0.01). During outpatient follow-up, bulging anterior fontanel, progressive macrocephaly, and enlarging ventricles each demonstrated a positive predictive value (PPV) of no less than 71%, but a bulging anterior fontanel remained the most predictive indicator of ETV/CPC failure, with a PPV of 100%, negative predictive value of 73%, and sensitivity of 72%. The highest PPVs and specificities existed when the clinical metrics were present in combination, although sensitivities decreased expectedly. Only 48% of failures were diagnosed on the basis all 3 hydrocephalus metrics, while only 37% of successes were negative for all 3 metrics. In the remaining 57% of patients, a diagnosis of success or failure was made in the presence of discordant data.
Successful ETV/CPC for infantile hydrocephalus was evaluated in relation to fontanel status, head growth, and change in ventricular size. In most patients, a designation of failure or success was made in the setting of discordant data.