Idiopathic Normal Pressure Hydrocephalus: New Findings and Thoughts on Etiology

Having been interested in normal pressure hydrocephalus (NPH) for a quarter of a century, I am gratified to see 2 articles on this topic in this issue of the American Journal of Neuroradiology (AJNR).^1,2 Because of the original description of NPH by Adams et al³ in 1965, many patients were shunted with only the symptom of dementia and, naturally, did not do well. Many questioned whether the disease even existed in the mid 1970s.^4,5 Fast forward 30 years to an editorial by neurosurgeon Robert Spetzler (Director of the Barrow Neurologic Institute),⁶ who stated that NPH may account for as many as 10% of cases of dementia.

In the current issue of AJNR, Antonio Scollato et al (also a neurosurgeon) report on a series of patients diagnosed clinically with NPH who refused ventriculoperitoneal shunt surgery.¹ He performed MR phase-contrast CSF flow studies on them every 6 months for the next 2 years and discovered something very interesting: In some patients, the aqueductal CSF stroke volume (ACSV) increased on follow-up without any treatment. More than 10 years ago, I wrote an article indicating that if the ACSV was not elevated, the patients had less chance of responding to shunt surgery.⁷ Specifically, the positive predictive value of shunt response for an ACSV >42 μL was 100%, whereas for stroke volumes less than 42 μL, it was 50%.

The way I interpret Scollato's findings is that the ventricles continue to enlarge after the patients become symptomatic with NPH. During the period before central atrophy sets in, the systolic expansion of the brain pushes against a larger drumhead, increasing the ACSV. Thus there will be a peak in the ACSV-versus-time curve when the ventricles reach their maximal expansion before atrophy (with decreased systolic expansion of the brain). I had always assumed that the 50% of patients with ACSV <42 μL who responded to shunt surgery had “some” atrophy; now I realize that we might have studied them too early in the course of their disease and that their ACSVs might have subsequently increased, putting them into the “hyperdynamic CSF flow” mode.

As noted by Scollato, the number 42 μL is machine-dependent and could vary considerably from manufacturer to manufacturer and even from software level to software level. Thus, I recommend that anyone doing these CSF flow studies perform 10 studies on elderly patients without symptoms of NPH and without dilated ventricles to determine what is normal. I make a diagnosis of “hyperdynamic CSF flow” (which supports a diagnosis of shunt-responsive NPH) when the ACSV is twice normal.

Also in this issue of AJNR, there is an article by radiologist Grant Bateman, arguing that deep white matter ischemia (DWMI) is not a cause of NPH but rather due to decreased compliance of the superficial veins.² Using an elegant phase-contrast technique that measures arterial inflow, deep and superficial venous outflow, and ACSV, he shows that patients with cerebral blood flow higher than average have normal drainage from the straight sinus but 9% less drainage from the superior sagittal sinus compared with age-matched controls. I am not convinced that normal flow in the straight sinus proves that there is a lack of deep white matter ischemia, given the many articles that now document its increased prevalence in patients with NPH.⁸ Although he does not question that deep white matter ischemia is present, he suggests that it is an epiphenomenon rather than the cause of NPH. Given the outward expansion of the brain against the inner table of the calvarium in communicating hydrocephalus, one might question whether the slightly decreased flow in the superior sagittal sinus is due to mild compression of the superior sagittal sinus and superficial cortical veins. Bateman notes the decreased compliance of the brain, based on decrease in the arteriovenous delay compared with healthy individuals. Could this also be an epiphenomenon rather than the cause of NPH? What causes some elderly patients to have decreased compliance in the first place?

Personally, I believe that DWMI is a cause of NPH but not the only cause.⁹ I believe that NPH starts in infancy as benign external hydrocephalus also known as “benign macrocrania of infancy.” This is a process that occurs in infants who present with an increasing percentile of head circumference compared with body length or weight. It has always been considered to be due to decreased resorption of CSF by “immature” arachnoidal granulations, which cannot keep up with the production of CSF. Because the sutures are still open at that age, the CSF accumulates in the slightly enlarged ventricles and in the frontal subarachnoid space and the head enlarges. We neuroradiologists caution against shunt surgery in these children because the arachnoidal granulations must mature at some point.

Actually, I am not sure they do. I think these individuals will always have decreased CSF resorption. Because saline infusion tests are not performed on babies, no one has really ever shown that CSF resorption improves. I think some of these babies will develop NPH 70 years later when they have a “second hit,” namely, DWMI.⁹ But let us go back and fill in some blanks.

With decreased CSF resorption via the arachnoidal villi and granulations, CSF needs a parallel path to exit the ventricles. I believe that this is the same path as is used by children with tectal gliomas, namely via the extracellular space of the brain, eventually punching through the pia into the subarachnoid space.⁹ This can be modeled as a parallel electrical circuit diagram, corresponding to CSF outflow via the foramina of Lushka and Magendie and via the extracellular space of the brain.⁹ Everything stays in balance (albeit with mild ventricular enlargement) until the onset of DWMI 70 years later when the symptoms of NPH appear.

DWMI is generally considered to be due to slow occlusion of the medium-sized vessels supplying the deep white matter, leading to the slow death of oligodendroglia. (This is not likely to be of sufficient magnitude to cause decreased blood flow in the straight sinus.) Histopathologically, DWMI appears as myelin pallor,¹⁰ (ie, less lipid and more water, which is why it is bright on a T2-weighted images). Regarding the 70 year olds who had benign external hydrocephalus as infants, the CSF in the extracellular space, which was previously gliding over the myelin lipid, is now being attracted to the naked myelin protein. This increases resistance to CSF flow through the extracellular space, leading to back up of CSF, more ventricular enlargement, and symptoms of NPH.⁹

Is there any evidence for this theory? When more than 50 patients with clinical NPH and hyperdynamic CSF flow (ACSV 3–4 times normal) were compared with sex-matched controls, their intracranial volumes were significantly larger (P < .003 for men and P < .002 for women). The difference in volumes was only approximately 100 mL; however, it definitely suggested that the process began in infancy when the sutures could enlarge.¹¹ When apparent diffusion coefficient measurements were compared in patients with NPH and age-matched controls, they were significantly higher in the periventricular white matter for a given degree of DWMI, consistent with outwardly flowing CSF being dammed up by the DWMI.⁹ I also have seen a number of cases in which the ventricles were enlarged 20 years before the patient developed symptoms of NPH, again suggesting that the process began much earlier.¹¹ How many times have we neuroradiologists seen enlarged ventricles without an obvious cause? We say “ventricles at the upper limits of normal” or “mild ventricular enlargement of uncertain etiology.” These patients may develop NPH in the future, and as Scollato points out, there is only a limited temporal window to treat them.