Table. Results of Neurological Examination Scores and Indices o f Motor, Sensoy, and Autonomic Neuropatby

Increment with Respect to Values at Entry

Entry 1 Year 2 Years 3.5 Years 5 Years 7 Years 10 Years

Immunosuppressed (n) 30 16 I5 14 11 10 NEx score -21.3 i 17.8 -2.70 t 9.48“ -3.00 t 9.04 -7.83 f 3.87‘ -11.0 f 6.26“ -15.0 t 8.81” -19.2 i 13.2”

MNC index -2.41 2 1.53 -0.28 -C 0.57’ -0.39 t 0.70’ -0.54 2 0.66“ -0.60 f 0.74“ -0.90 f 1.05” -1.78 f l . l l b

SNC index -3.10 i 2.05 -0.37 t 0.93” -0.47 2 0.79” -0.47 f 1.02 -0.56 t 0.94“ -0.85 i 0 . W -1.32? 1.14“

(4 12 6 3 3 1 1

CRR index -2.04 f 0.52 -0.15 t 0.19” -0.42 -C 0.65 -0.23 t 0.96 -0.41 t 0.88 -0.59 -0.65 Not immuno-

suppressed (n) 47 14 6 5 3 1

NEx score -27.4 f 14.4 -4.20 f 6.81b -5.33 f 4.58‘ -6.40 f 4.16“ -7.75 t 7.61‘ -7.50 t 7.71”

MNC index -2.12 t 1.67 -0.42 f 0.67b -0.44 t 0.63” -0.67 t 0.94^ -0.55 t 0.84 -0.34 t 0.35

SNC index -3.10 t 1.84 -0.56 t 1.0Ib -0.93 2 1.38” -0.69 t 1.09 -0.66 t 0.87 -0.29 f 0.19

(n) 44 13 5 4 2 CRR index -1.77 t 0.85 -0.19 t 0.53” -0.30 -C 0.98 -0.17 C 0.73 -0.08 t 0.63 -0.12 t 1.23

In diabetic patients of the control yp with 9 renal transplant (immunosuppressed) and without any transplant or a failed pancreatic transplant alone (not immunosup- pressed) at entry into the study an

‘p < 0.05; ‘p < 0.005, vs values at entry of the same subgroup. There were no significant differences between the two subgroups for any measurement.

NEX = neurological examinations; MNC = motor nerve conduction; SNC = sensory neme conduction; CRR = cardiorespiratory reflexes.

their variations over 10 years of follow-up. Data are mean 2 SD values.

cessful pancreatic tran~plantation’,~ may be mostly attribut- able to the restoration of a normoglycemic state.

*Department of Cell Biology and Pbysiolog, Universitat Autdnoma de Barcelona, Bellaterra, Spain; and ?Department o f Neurology, University of Minnesota, Minneapolis, M N

References 1. Navarro X, Sutherland DER, Kennedy WR. Long-term effects o f

pancreatic transplantation on diabetic neuropathy. Ann Neurol

2. Krendel DA, Costigan DA, Hopkins LC. Successful treatment o f neuropathy in patients with diabetes mellitus. Arch Neurol 1995; 52:1053-1061

3. Van der Vliet JA, Navarro X, Kennedy WR, et al. Long term follow-up of polyneuropathy in diabetic kidney transplant recip- ients. Diabetes 1988;37: 1247-1252

4. Najarian JS, Kaufman DB, Fryd DS, et al. Long-term survival following kidney transplantation in 100 type I diabetic patients. Transplantation 1989;47:106-113

5. Kennedy WR, Navarro X, Goetz FC, et al. Effects o f pancreatic transplantation on diabetic neuropathy. N Engl J Med 1990; 322: 1031-1037

1997;42:727-736

Viral Load in HIV-Associated Dementia Jonathan D. Glass, MD,* and Steven L. Wesselingh, MBBS, PhDt

In their recent report in Annals, Ellis and colleagues’ ad- dressed the very important issue of correlation between cere- brospinal fluid (CSF) “viral l o a d and cognitive impairment, and concluded that human immunodeficiency virus (HIV) RNA levels in CSF directly correlated with “neuropsycholog- ical impairment” and that CSF viral measurements are useful for monitoring patients during therapy. The data presented, however, are less than convincing. Of the 97 patients stud- ied, 40 were designated “neuropsychologically impaired,” in-

cluding 27 with minor cognitive motor disorder (MCMD) and 3 with HIV-associated dementia. Twenty-eight of these 40 had CD4 counts of 200 or higher and showed no statis- tical difference in CSF RNA levels when compared with cog- nitively normal patients. Twenty-one of these 28 had MCMD or HIV-associated dementia. Statistical significance was only shown for those with CD4 counts of less than 200, a group with only 12 patients, 10 of whom had MCMD or HIV-associated dementia. Thus, 70% (28 of 40) of the cog- nitively impaired patients showed no difference in CSF RNA levels.

Only 4 patients overall (5%) were diagnosed with HIV- associated dementia, 3 in the group with less than 200 CD4 cells (the only group showing significant differences in CSF RNA levels). There is no demonstration that these patients had higher levels than the 9 with neuropsychological impair- ment or MCMD. If present, this positive “dose-response’’ data would be supportive of the authors’ conclusions. We suggest, also, that grouping patients with neuropsychological testing abnormalities only (“neuropsychologically impaired) with those with HIV-associated dementia is unjustified, be- cause there is no indication that these patients have the same disease process or will progress to HIV-associated dementia.

In summary, the data presented in this article are not sub- stantial enough to support a direct correlation between CSF viral load and HIV-associated dementia, and are more sup- portive of a direct relationship between CD4 count and CSF viral load. The search for a direct link between brain viral load and HIV-associated dementia has thus far uncovered only weak correlations. We believe that evidence in the lit- erature, to date, better supports an indirect role for HIV in the pathogenesis of HIV-associated dementia,2 and supports our previous statement that viral presence in the nervous sys- tem is necessary but not sufficient for the development of HIV-associated dementia.3

150 Annals of Neurology Vol 44 No 1 July 1998

*Emory University, Atlanta, GA; and fFlinders University, Bedford Park, South Australia, Australia

References 1.

2.

3.

Ellis R, Hsia K, Spector S , et al. Cerebrospinal fluid human im- munodeficiency virus type 1 RNA levels are elevated in neuro- cognitively impaired individuals with acquired immunodeficiency syndrome. Ann Neurol 1997;42:679-688 Epstein LG, Gendelman HE. Human immunodeficiency virus type 1 infection of the nervous system: pathogenetic mecha- nisms. Ann Neurol 1993;33:429-436 Glass JD, Fedor H, Wesselingh SL, McArthur JC. Immunocy- tochemical quantitation of human immunodeficiency virus in the brain: correlations with dementia. Ann Neurol 1995;38: 755-762

Reply Ronald J. Ellis, MD, PhD, and J. Allen McCutchan, MD

We appreciate the careful reading of our article undertaken by Drs Glass and Wesselingh. They correctly summarize the data from our prospective cohort study' but overlook two important points of emphasis.

The title of the article summarizes one of these points, which is that among subjects with acquired immunodefi- ciency syndrome (AIDS), cerebrospinal fluid (CSF) human immunodeficiency virus type 1 (HIV-1) RNA levels were el- evated in those who showed neuropsychological impairment compared with those who did not. The same was not true for HIV-infected subjects who did not have AIDS. This is not a trivial distinction. Individuals with AIDS, defined by the Centers for Disease Control in 1993 as those with a CD4+ lymphocyte count of less than 200, have a dramati- cally increased risk of major medical and neurological mor- bidities and mortality compared with those without AIDS. Using data provided in the published study, the calculated odds ratio for the association between CSF HIV RNA levels greater than 200 and neuropsychological impairment in AIDS is 19.

Because the preponderance of severe neurological disease occurs in patients with AIDS, and because our observations suggest that elevated CSF HIV RNA levels are strongly as- sociated with neuropsychological impairment (including HIV-associated dementia and minor cognitive motor disor- der) in AIDS, these findings are of considerable potential clinical importance. If prospective studies demonstrate that lowering elevated CSF HIV RNA levels with carefully cho- sen antiretroviral agents can benefit patients' cognitive status, then a long-sought goal of HIV clinical neurological research will have been achieved.

The second point deals with a potential explanation for the lack of association between CSF HIV RNA levels and neuropsychological impairment in subjects who did not have AIDS. We found that CSF pleocytosis (elevated leukocyte numbers) was common before AIDS, and that subjects with pleocytosis typically had CSF HIV RNA levels much closer to those in blood plasma than did subjects without pleocy- tosis. These data suggest that elevated CSF HIV RNA levels occur for different reasons before and after AIDS. The find- ings are consistent with the notion that an increase in cellu-

lar trafficking between the blood and the CSF in some pa- tients before AIDS may result in higher HIV RNA levels in the CSF compartment. As clearly shown in Figure 2, there was no direct relationship between CD4 count and CSF HIV RNA.

We agree that the preponderance of data accumulated thus far suggest that multiple pathophysiological mechanisms contribute to the development of HIV-associated dementia. Our findings suggest that the presence of virus in the CSF in late disease may be a marker of virus in other CNS tissues, and that appropriate treatment may interrupt the subsequent cascade of events that leads to neuronal injury and clinical neurological disease.

Departments o f Neurosciences and Medicine, University of California, San Diego, C 2

Olivopontocerebellar Atrophy and Multiple System Atrophy: Clinical Follow-Up of 10 Patients Studied with PET A. Schrag, MD,* J. 0. Rinne, M D , t D. J. Burn, MD, MRCP,$ C. J. Mathias, DPhil, FRCP,* C. D. Marsden, MD, FRS, DSc,* D. J. Brooks, MD, FRCP,$ and N. P. Quinn, MD, FRCP*

Many patients with sporadic olivopontocerebellar atrophy (sOPCA) develop other symptoms and signs to indicate that they have multiple system atrophy (MSA).' Three years ago, in Annals, we reported ['*F]fluorodopa and ["Cldiprenor- phine positron emission tomography (PET) data in 10 pa- tients with the cerebellar form of MSA with autonomic symptoms (8 severe and 2 mild), to clarify what proportion of these patients also have nigrostriatal dysfunction.2 In 7 patients, striatal tracer uptake was significantly reduced even in the absence of clinical parkinsonism in 5 of them, indi- cating that PET can detect subclinical nigrostriatal dysfunc- tion in sOPCA. An eighth patient with mild finger akinesia had a fluorodopa K, value reduced 1 SD below the control mean, and the other 2 patients had normal PET scans and no parkinsonism clinically. However, it remained uncertain what proportion of patients with normal or abnormal nigro- striatal function on PET scanning would later develop par- kinsonism. Here we report clinical follow-up of these 10 pa- tients (Table).

Now, 4 years after the examination, 7 patients have died (necropsy was, unfortunately, not obtained), 1 has emigrated, and 2 are still alive. All of the 7 patients with reduced flu- orodopa uptake developed additional features (pyramidal signs [n = 71, urinary incontinence or retention [n = 61, abnormal sphincter electromyogram (EMG) [n = 6 ] ) , con- firming OPCA-type MSA. In addition to the 2 who already had definite parkinsonism, a further 4 of these 7 patients went on to develop it; the seventh rapidly deteriorated to being wheelchair bound and died 6 months after his PET scan, but we do not know whether he developed clear signs of parkinsonism. The patient with slightly reduced fluoro- dopa uptake later became more parkinsonian. In the remain-

Annals of Neurology Vol 44 No 1 July 1998 151