Cerebrospinal fluid neopterin decay characteristics after initiation of antiretroviral therapy

In this retrospective study, we used stored CSF and blood specimens from two clinical centers in Gothenburg, Sweden, and San Francisco, USA. All patients who had undergone at least one lumbar puncture prior to initiation of combination ART and at least one lumbar puncture on treatment were considered for this study. Participants were either ART-naive or had been off treatment for ≥6 months at ART initiation (baseline). Because this study focused on neopterin concentrations among virologically suppressed patients, only subjects with viral suppression after initiation of ART were included. Virological suppression was defined as CSF and plasma HIV RNA levels <50 copies/mL within 6 months from the start of therapy, or if followed for a shorter period, >2 log₁₀ reduction of the viral load. Patients with virologic failure, defined as two consecutive plasma samples with HIV RNA >50 copies/mL after 6 months of ART, were excluded. Decay and set-point models were based on a group of neurologically asymptomatic individuals fulfilling the above criteria. A small group of patients with HAD was also studied for comparison. All subjects were studied within study protocols approved by the Research Ethics Committee of the University of Gothenburg or the University of California San Francisco Committee on Human Research.

Neopterin was measured by a commercially available radio-immunoassay or enzyme-linked immunosorbent assay (Neopterin RIA and EIA, BRAHMS, Hennigsdorf, Germany) [13]. The same antibodies are used in these two assays, and the results are interchangeable [14]. Normal neopterin reference values were <8.8 nmol/L in blood and <5.8 nmol/L in CSF [4, 13, 15]. HIV-1 RNA in CSF and plasma was measured using the Roche Amplicor assay (version 1.5, Hoffman-La Roche, Basel, Switzerland). The assay has a lower detection limit of 20 copies/mL (1.30 log₁₀ copies/mL). All HIV-1 RNA values <20 copies/mL were set as 19 copies/mL. The CD4⁺ T-cell count was analyzed by flow cytometry in local clinical laboratories.

Descriptive statistics were generated for all variables involved in the analyses. Comparisons of continuous statistical measures between the non-HAD and HAD patients were performed using the Kruskal-Wallis test. We used non-linear mixed models to reflect the decay characteristics of both CSF and blood neopterin levels after initiation of ART. From previously published reports [16, 17] we assumed that CSF and blood neopterin levels would decay rapidly after initiation of treatment, and would stabilize after some time reaching a set-point. The model used for both the CSF and blood neopterin levels is given by the following equation:

where y _ij corresponds to neopterin levels for each subject and post-baseline study visit, t _ij, X _i is the log-transformed pre-treatment blood or CSF (log-transformed) neopterin levels, β ₀ , β ₁ , govern the association between follow-up CSF or blood neopterin levels overall and with baseline levels respectively, β ₂ and β ₃ estimate the reduction in CSF or blood neopterin levels from baseline to the set-point levels after prolonged treatment, overall and associated with baseline levels, and β ₄ and β ₅ determine the rate of decay overall and according to the pre-treatment neopterin levels respectively. Subject-specific random effects u _0i and u _2i were added to allow initial and set-point neopterin levels to vary for each patient. The impact of the pre-treatment neopterin level on the post-baseline visits, the ultimate set-point reached, and the rate of decay of neopterin levels were tested using the Wald test, involving the coefficients β ₁, β ₃, and β ₅ respectively. The size and sign of the model coefficients is interpreted as follows: an estimate β ₁ >0 implies a positive association between pre-treatment and post-treatment neopterin levels. The estimated set-point is (β ₀ + β ₂ ) + (β ₁ + β ₃ )X _i . This suggests that an estimate β ₁ + β ₃ >0 implies that patients with higher pre-treatment CSF neopterin levels would end up with a higher set-point, while if β ₁ + β ₃ <0 the exact opposite is concluded. In addition, β ₅ >0 implies a decay rate associated with baseline CSF neopterin levels that is faster than the average rate of decay, which is estimated by β ₄, while β ₅ <0 suggests that higher baseline neopterin levels are associated with slower rates of decay. Other models involving factors potentially associated with long-term neopterin levels were also considered. They included longitudinally measured HIV RNA levels in the CSF and plasma and CD4⁺ T-cell counts at the start of therapy. In the presence of pre-treatment neopterin levels, no other factors were found to be significantly associated with long-term CSF or blood neopterin levels, so the model described above was used.

In addition to producing point estimates of the neopterin set-point for each subject, we tested whether the estimated set-point corresponding to the median and third quartile levels of baseline CSF and blood neopterin in the neurologically asymptomatic population was significantly higher than the upper limit of normal using the Wald chi-square test. This test compares the difference between the upper limit of normal from the estimated neopterin set-point scaled by the variance of the estimate, and provides a criterion for deciding whether this difference is large enough to be considered statistically significant. Using the model developed for the neurologically asymptomatic patients, we also tested whether the estimated CSF and blood neopterin set-point corresponding to the median and third quartile levels of baseline CSF and blood neopterin among subjects with HAD was significantly higher than the upper limit of normal of CSF and blood neopterin levels in the same manner. All analyses were performed with R version 2.11.1 (The R Foundation for Statistical Computing).

In total, 102 chronically HIV-1-infected individuals without neurological symptoms fulfilled the inclusion criteria and were included in the study (88 from Gothenburg and 14 from San Francisco) between 1996 and 2009. All neurologically asymptomatic subjects were initiated on combination ART with a median CNS penetration effectiveness (CPE) score of 8 (range 5 to 13) [18]. As a comparison group, seven subjects with HAD (one from Gothenburg and six from San Francisco) were analyzed separately and compared to the non-HAD group. They initiated ART with a median CPE score 9 (range 6 to 13). Subject baseline characteristics are presented in Table 1. Baseline was defined as the closest time point prior to ART initiation. This ranged from 0 to 36 weeks prior to the start of therapy, with a median value of 0.7 weeks. The number of evaluations and the extent of follow-up varied among the subjects. In the non-HAD group median follow-up was 84.7 months compared to 21.6 months among the HAD subjects (log-rank P = 0.017). The two patient groups were similar with respect to age, sex, CD4⁺ T-cell count, and plasma HIV RNA levels at baseline. CSF HIV RNA and neopterin levels were significantly higher among HAD-patients than non-HAD subjects (P = 0.001).

The neurologically asymptomatic subjects stayed asymptomatic during the whole study period, but structured neuropsychological testing was not performed. All seven HAD-patients improved functionally after initiation of ART. Neuropsychological testing was performed in six HAD-patients. Four of these improved substantially. One got worse and one remained about the same on testing. The average change in NPZ-4 NPZ (composite neuropsychological test Z score) score was 1.8 (improved) with SD of 2.3.

The median pre-treatment CSF neopterin level for the 102 neuroasymptomatic subjects was 19.9 nmol/L (range 4.0 to 138.0). Individual and overall average CSF neopterin decay trajectories for these patients are presented in Figure 1. Forty-two out of the 102 neuroasymptomatic patients (41%) had estimated set-point CSF neopterin levels that were higher than the upper limit of normal (5.8 nmol/L). The estimated model of CSF neopterin decay was:

All coefficients were significantly different from zero (P-value <0.0001 in all cases). The model (see Methods) is interpreted as follows: subjects with higher pre-treatment CSF neopterin levels were expected to have higher set-points compared to subjects with lower baseline levels because of the positive sum of pre-treatment and post-treatment coefficients (β ₁ + β ₃=3.3). This means that a subject with a one-log higher pre-treatment CSF neopterin level compared to another subject would be expected to have a set-point 3.3 nmol/mL higher. Conversely, subjects with pre-treatment CSF neopterin at the third quartile (33 nmol/mL) are estimated to have a 1.72 nmol/mL higher set-point compared to subjects with pre-treatment CSF neopterin levels near the median (19.6 nmol/mL). In addition, subjects with higher pre-treatment CSF neopterin levels experienced faster rates of decay than subjects with lower pre-treatment CSF neopterin levels as suggested by the positive sign of coefficient β ₅ >0 There was no association between pre-treatment CD4⁺ T-cell count or HIV RNA levels in CSF or blood with follow-up CSF neopterin levels, so these factors were excluded from any further consideration in the analyses.

The median pre-treatment blood neopterin level in non-HAD subjects was 21.7 nmol/L (range 6.8 to 77.0) (Table 1). The median blood neopterin level among HAD subjects was 25.0 nmol/L (range 12.0 to 28.9), which was not significantly different from non-HAD patients (P = 0.359). An identical analysis to the one presented for CSF neopterin levels was carried out with respect to blood neopterin levels. The model fitted was:

Pre-treatment blood neopterin levels constituted the only factor significantly associated with post-treatment levels (P <0.0001). Subjects having higher pre-treatment levels were estimated to have higher set-point levels as implied by the sum of the coefficients β ₁ + β ₃ = 0.78. In contrast to the CSF neopterin model, baseline blood neopterin levels were weakly associated with the rate of decay after initiation of ART estimated by β ₅ (P = 0.091). That is, blood neopterin levels decreased in uniform fashion for all patients with those having the highest blood neopterin levels decreasing slightly more slowly.

It was estimated that 28 out of 102 (27%) non-HAD subjects and 6 out of 7 (86%) HAD-subjects had blood neopterin set-point levels above the upper limit of normal. This is seen from the individual and overall trajectories presented in Figure 2.