Lumboperitoneal shunt for the treatment of pseudotumor cerebri

Pseudotumor cerebri (PTC), also called benign intracranial hypertension (BIH), is characterized by increased intracranial pressure unassociated with evidence of an intracranial mass, enlargement of the ventricles, or abnormal composition of the cerebrospinal fluid (CSF). The cause of PTC in most patients is unknown but thought to be related to decreased outflow of CSF. ^[1]

The most significant sequela of PTC is permanent loss of vision caused by prolonged papilledema with secondary optic atrophy. Although evidence of visual sensory loss is demonstrable in up to 90% of patients with PTC, ^[2,3] most of these patients have only minor defects in the visual fields, particularly enlargement of the physiologic blind spots. With time, however, patients may experience progressive constriction of the visual fields, disturbances in color vision and, eventually, loss of central vision. ^[4]

The initial treatment of PTC usually is with drugs such as acetazolamide that reduce intracranial pressure ^[5]; however, the presence of significant optic neuropathy at presentation or the development or progression of optic neuropathy despite maximum medical therapy generally requires surgical intervention. ^[6] Surgery also may be indicated in patients with PTC who have intractable headache.

The two primary surgical options for patients with PTC are optic nerve sheath fenestration (ONSF) and placement of a lumboperitoneal shunt (LPS). LPS was once considered the optimum surgical treatment for PTC; however, recent reports have suggested that ONSF is more effective and associated with fewer complications than LPS. ^[7-10] Because of these reports, many physicians have abandoned LPS in favor of ONSF for the majority of their patients with PTC who require surgery. Nevertheless, long-term follow-up data suggest that ONSF may not be as effective as originally claimed, ^[11] and numerous complications of ONSF have been reported as more physicians perform the procedure. ^[12-16] Because of increasing questions regarding the safety and longterm efficacy of ONSF, we retrospectively studied a population of patients with PTC who were treated at The Johns Hopkins Hospital with at least one LPS to assess the long-term effectiveness and complication rate of LPS in this condition.

Methods.

The medical records of all patients with PTC who were evaluated in the Neuro-Ophthalmology Unit of The Johns Hopkins Medical Institutions between 1975 and 1993 and who subsequently were treated with a LPS were reviewed. Inclusion criteria for analysis included diagnosis of PTC by a combination of neuroimaging studies, measurement of intracranial pressure, and analysis of CSF. Patient age, sex, symptoms, and body habitus were recorded, as were the results of the neuro-ophthalmologic examination, indications for LPS, and subsequent course. Actual patient weights were recorded when available in the records; otherwise, a physical description of the body habitus as documented by the examining physician was noted. When our most recent neuro-ophthalmologic evaluation was earlier than 1994, patients and their physicians were contacted to obtain more recent data. The cumulative risk of revision after LPS was calculated by using standard survival analysis methods. Survival curves for different risk subgroups were compared by using the log rank test.

Results.

Twenty-seven patients fulfilled the above criteria and were included in the analysis Table 1. This group included 24 females and three males. Twenty-six patients (96.2%) had no structural lesion identified as the cause of their PTC and were thus classified as having idiopathic pseudotumor cerebri. One patient (no. 1), an 8-year-old boy, developed PTC in the setting of acute left-sided mastoiditis with secondary occlusion of the ipsilateral transverse sinus. The mean age of all patients at diagnosis was 28 years (median, 29 years). The mean age of all patients at time of the initial shunt was 30 years (median, 32 years).

Intractable headache was the most common indication for placement of a LPS in our series, being cited as at least one reason in 18 of the 27 patients (67%). All of these patients experienced relief of, or improvement in, cephalalgia after the LPS.

Fourteen patients (52%) underwent LPS for progressive optic neuropathy related to persistent elevated intracranial pressure. In nine of these patients, the optic neuropathy alone was the indication for the LPS, whereas in the other five patients, headache was also an indication for surgery. In no case was failure of the LPS responsible for acute visual loss. Visual function, as measured by testing of visual acuity, color vision, and either kinetic or static visual fields, returned to normal in both eyes of six patients (nos. 1, 6, 8, 10, 20, and 24), showed no change in either eye in four patients (nos. 7, 17, 18, 19, and 24), and improved in at least one eye in the remaining four patients (nos. 11, 15, 23, and 24). Improvement was defined as an improvement of visual acuity of at least one line of best-corrected Snellen acuity, improvement in color vision of at least one full plate as measured with Hardy-Rand-Rittler pseudoisochromatic plates, improvement in static perimetry by at least 1 dB of mean deviation, improvement of 5 degrees or more in at least one quadrant of the peripheral field to at least two stimuli by kinetic perimetry, or a combination of these findings.

Four patients had a unilateral abducens nerve paresis, and one patient had a bilateral abducens nerve paresis at the time of LPS. In none of these patients, however, was the paresis an indication for the surgery. All five patients experienced complete resolution of the paresis after LPS.

All 24 women in this study were considered obese by virtue of their weight and height by their examining physicians; however, an accurate weight was recorded for only 13 of them (54%). The average weight of these patients was 254 pounds. None of the three males in the study was described as obese. One of these patients was the 8-year-old boy (patient 1) in whom PTC was caused by septic occlusion of the left transverse sinus related to otitis media. In neither of the other two males could a cause of PTC be identified, despite extensive assessments in both cases.

The duration of follow-up ranged from 21 months to more than 23 years, with a mean of 77 months and a median of 47 months. During the period of follow-up, 12 patients (44%) required no revision, whereas 15 patients (56%) required a total of 66 revisions. Among the 15 patients who required shunt revision, 6 (40%) required one revision and 2 (13%) required two revisions. The remaining 7 patients (47%) required more than two revisions, with three of them requiring a total of 35 revisions (patient 5, 13 revisions; patient 13, 12 revisions; patient 15, 10 revisions) or 53% of the total number of revisions performed.

When one considers all 27 patients in the study, the average number of revisions per patient was 2.4, with one revision being performed per 31.5 shunt-months or 2.6 shunt-years, and 0.38 shunt revisions being performed per year. When only the 15 patients who actually required a shunt revision are considered, however, the average number of revisions was 4.4, with one revision performed per 19.7 shunt-months or 1.6 shunt-years, and 0.61 revisions being performed per year of follow-up.

The most common reason for LPS revision was shunt obstruction with recurrence of symptoms and signs of increased intracranial pressure (43 of the 66 procedures or 65% of the total number of revisions). Other reasons for shunt revision were valve malfunction with development of low-pressure symptoms (10 revisions or 15.1% of total), catheter migration or dislocation (3 revisions; 4.5% of total), radiculopathy (3 revisions; 4.5% of total), shunt-related infection (1 revision; 1.5% of total), and abdominal pain (1 revision; 1.5% of total).

Patients who required a shunt revision (n = 15) tended to be slightly younger at the time of diagnosis of PTC (24.9 years versus 32.9 years), to have longer follow-up (86.5 months or 7.2 years versus 65.0 months or 5.4 years), and to be heavier (where weights were recorded--268.5 lbs [n = 8] versus 232.2 lbs [n = 5]) than patients who did not require a shunt revision. We did not have adequate sample size to formally test for differences in risk associated with these factors; however, no patient whose diagnosis of PTC was made when he or she was 35 years of age or older (n = 6) required a shunt revision.

The risk of a shunt revision was calculated by using a survival analysis approach. This approach allows the use of data from all patients in the series, whether or not they required a revision, and it adjusts for follow-up time. The cumulative probability of a revision at each month of follow-up is shown in Figure 1. The median time to revision for this series was 11 months, with only 1 of the 15 patients (no. 5) having a first revision performed more than 12 months after placement of the initial LPS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Cumulative risk of initial revision of lumboperitoneal shunt (n = 27).

We also calculated the cumulative probability of a second shunt revision, once a first revision had been performed Figure 2. The median time to second revision after first revision was 10 months.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Cumulative risk of second revision after first revision has occurred (n = 15).

Finally, we recalculated all statistics when only the single case of nonidiopathic PTC (patient 1) was excluded from consideration and also when all three nonobese men were excluded from the study. In neither setting did these recalculations substantially change the mean or median age of the patients at diagnosis or at first shunt, the percentage of patients requiring shunt revision, the percentage of patients requiring only one or two revisions, or the percentage of patients requiring more than two revisions. In addition, the calculations did not alter the median times to first or second revision.

Discussion.

A number of reports have documented both the efficacy and complication rate of LPS in patients with PTC. ^[8,9,17-23] Although placement of a lumboperitoneal shunt is generally safe, any operative procedure performed under general anesthesia carries some risk, and there is at least one report of a perioperative death following LPS placement. ^[24] We had no perioperative deaths in our series, nor was there any perioperative morbidity. In addition, none of the patients with a functioning shunt had any shunt-related symptoms, such as low-pressure headaches or abdominal pain.

Obstruction is the most common complication of LPS in most large reviews in the literature, ^[19,22,23] and it also was the most common complication in our series, being responsible for 65% of revisions. PTC is sometimes considered a self-limited disease. ^[25] In fact, many patients whose symptoms and signs improve without treatment, and others who improve on medical therapy and then remain asymptomatic after therapy is stopped, nevertheless have persistent increased intracranial pressure when tested by lumbar puncture. ^[26] Thus, rather than being self-limited, PTC tends to be a chronic disease. Indeed, Repka et al. ^[27] reported several patients with PTC successfully treated by LPS who developed acute visual loss from ischemic optic neuropathy caused by recurrent increased intracranial pressure following elective removal of the shunt several years after placement, at a time the disorder was thought to have resolved. Although patient 6 was one of the patients reported by Repka et al., ^[27] neither her initial LPS nor her subsequent five revisions were performed for acute visual loss, and none of the other patients in this series experienced acute visual loss from shunt dysfunction.

Secondary intracranial hypotension caused by excessive drainage of CSF via the LPS was the second most common complication leading to shunt revision in our series. Ten revisions, or 15.1% of the total number of revisions, were performed to alleviate signs and symptoms caused by this complication. Despite pressure valves to prevent this complication, low-pressure symptoms account for a significant proportion of LPS revisions in most other series ^{[17,19,20,23,28-30]} and may be a particular problem in children with pseudotumor cerebri. ^[22]

Lumbar radiculopathy is a complication of LPS. ^[23,24,29] Five patients in our series complained of back pain caused by lumbar radiculopathy, three of whom required shunt revision for this symptom, accounting for 4.5% of all revisions.

Although shunt infection is a potentially disastrous complication of LPS, ^{[17,19,23,24,29-32]} it occurs in only about 1% of patients. ^[29,31] One shunt revision (1.5%) in our series was performed because of a shunt infection. Other infectious complications may also occur following LPS. Cabezudo et al. ^[32] reported the case of a patient with PTC who developed a disk space infection 6 weeks after insertion of an LPS. We encountered no such instances of other infections related to LPS in the patients in our series.

Complications related to the peritoneal end of the catheter can occur after lumboperitoneal shunting as with ventriculoperitoneal shunting. ^{[22,24,28,30,31,33-37]} There was one revision in our series (1.5%) because of shunt-related abdominal pain, and three revisions (4.5% of the total) because of migration or dislocation of the peritoneal end of the catheter.

Both tonsillar herniation and syringomyelia can occur after LPS, and these complications may be symptomatic or asymptomatic. ^[22,38-41] Sullivan et al. ^[40] reported tonsillar herniation associated with the development of a cervicothoracic syrinx that occurred 3 years after placement of a LPS in a patient with PTC. MRI documented resolution of the herniation after the shunt was clamped and a ventriculoatrial shunt was placed. Chumas et al. ^[22,41] retrospectively studied 143 children who had undergone LPS. The indication for LPS in this population was hydrocephalus in 81% and pseudotumor cerebri in 7%. Within this cohort, five patients required suboccipital decompression for symptomatic tonsillar herniation (4.2%), and one patient died from unsuspected and previously asymptomatic tonsillar herniation. Seventeen other asymptomatic patients underwent MRI based on suspicious CTs, and 12 of these patients (70.6%) had tonsillar herniation that ranged from 2 to 21 mm. The average time from LPS to symptomatic tonsillar herniation was 6.9 years. On the basis of these findings, Chumas et al. ^[41] recommended that MRI be performed at regular intervals in all patients after LPS. Although routine MRI was not performed in our patients, one of the patients (no. 5) developed asymptomatic tonsillar herniation down to the level of C-1, associated with syringomyelia at the C7-T1 level, and a second patient (no. 1) also developed mild tonsillar herniation. Although we elected not to remove or alter the shunt in either of these patients, we agree with Chumas et al. ^[41] that physicians should consider screening asymptomatic patients for the development of tonsillar herniation after LPS and that such patients may require removal of the shunt.

Although the risk of subdural hemorrhage is greater following ventriculoperitoneal shunting than after lumboperitoneal shunting, subdural hematoma may develop after LPS. ^[35,42] None of our patients developed a subdural hematoma after LPS.

The primary goal of therapy for PTC is lowering of the increased intracranial pressure. LPS achieves this goal in most patients, whereas ONSF has not been convincingly shown to affect intracranial pressure. ^[43-47] In addition, although ONSF may result in immediate improvement in, or resolution of, optic neuropathy in patients with PTC, about one third of patients undergoing unilateral ONSF will not experience relief of headaches. ^[7-9] Spoor and McHenry ^[11] evaluated the long-term results of ONSF in 75 eyes of 54 patients with PTC and found that 32% of ONSF failed within 39 months after surgery. Furthermore, only about 75% of ONSFs appear to be functioning 6 months after surgery, and the probability of a functioning ONSF steadily decreases thereafter such that 66% of ONSFs are functioning at 12 months, 55% at 3 years, 38% at 5 years, and 16% at 6 years after surgery. ^[11] Although the study by Spoor and McHenry ^[11] can be criticized on methodologic grounds because of its highly selective and retrospective nature (e.g., patients doing well are less likely to return for follow-up and, thus, might not be included in the study) and its definitions of worsening (deterioration in visual field was defined as a reduction of more than 2 dB of mean deviation), the study indicates that, like patients who receive a lumboperitoneal shunt, patients who undergo an ONSF for PTC require careful follow-up for many years. Although patients may be treated with a second ONSF after initial failure, ^[48] eyes that have more than one ONSF rarely stabilize or improve after surgery and are more likely to experience a significant vascular complication than are eyes that undergo a single ONSF. ^[14]

Our patient population could be separated into two groups. Group I was composed of 12 patients whose intracranial pressure was normalized with a single LPS. Group II was composed of 15 patients who underwent at least one shunt revision. Patients in group I tended to be older and to weigh less than patients in group II. Although neither of these features was statistically significant, no patient in this series whose diagnosis of PTC was made when the patient was 35 years of age or older required a shunt revision. We also found that it was common for the first shunt revision to occur within 2 to 3 months of the initial LPS (cumulative risk, 37%). Indeed, only one patient (no. 5) underwent her first shunt revision more than 1 year after the initial shunt. Thus, a patient with PTC who undergoes a LPS and who maintains a functioning shunt for more than 1 year has a lower risk of requiring a shunt revision over subsequent years.

Five of the 15 patients in group II required one shunt revision, and three additional patients in this group required two revisions. Thus, 53% of patients who required revision of their shunt required one or two revisions during the period of follow-up, which averaged 7 years. Among the remaining seven patients in group II who required at least one shunt revision, three required a total of 35 shunt revisions or 53% of the total number of 66 shunt revisions performed in this group of patients. These latter three patients had no demographic features or clinical manifestations that allowed them to be distinguished from the other 12 patients who required at least one shunt revision.

The results of this study allow us to make a number of observations regarding the efficacy of lumboperitoneal shunting for PTC. First, a functioning LPS is effective in alleviating the signs and symptoms of PTC and in preventing the visual complications of this disorder. Second, major complications of lumboperitoneal shunting other than failure of the shunt are rare. Third, patients with a functioning shunt rarely have persistent shunt-related symptoms, such as low-pressure headaches or abdominal pain. Fourth, a substantial percentage of patients who undergo placement of a lumboperitoneal shunt for PTC do not require a revision of the shunt. Fifth, most of the patients who require a shunt revision need only one or two revisions over many years. Finally, patients who ultimately require shunt revision often do so within a few months, and almost always within 1 year, of placement of the initial shunt. Because of our small sample size, we were unable to identify any factors associated with risk of at least one revision of the shunt, although patients in this study who did not require a shunt revision tended to be older than those who did, and no patient who was 35 years of age or older at the time the diagnosis of PTC was made needed a revision. Similarly, we were unable to identify any factors associated with risk of multiple revisions.

There is currently no ideal treatment for PTC. Both ONSF and LPS, the primary surgical treatments for PTC, have inherent advantages and disadvantages. The decision regarding which treatment to employ in a particular patient must be individualized. Regardless of which method is used, however, the clinician must remain vigilant for treatment failures that may occur months to years after an initially successful therapy. We agree with Wall ^[49] that a prospective, randomized, multicenter PTC treatment trial should be performed.