In Western countries the initial HIV epidemic tended to be concentrated in urban areas within the gay male population. By contrast, in China the HIV epidemic began in rural areas in association with blood product collection and use, before spreading to urban injection drug users (IDUs), among whom it has reached very high rates of infection in some areas (up to 44% - Zhao et al, 2006). The epidemic among Chinese former plasma donors (FPDs), on which this current study is focused, has been centered in agrarian provinces such as Anhui, Hebei, Henan, Hubei and Shanxi (Cohen, 2004). In these areas prior to the mid 1990s, a large number of poorly regulated commercial plasma collection companies were in operation, some using non-sterile techniques. Residents would supplement their small farming incomes by selling their plasma, in some cases very often, unknowingly placing themselves at risk for exposure to HIV and other blood borne pathogens. In the mid 1990s the Chinese government imposed strict regulations to prevent further spread of HIV in this way.

Estimates of future infections are difficult, but UNAIDS and other organizations have predicted that by 2010, there could be a generalized epidemic with between ten and twenty million HIV-infected Chinese (UNAIDS, 2006). Because China is the most populous country on earth, with 1.3 billion people, the growth of the HIV epidemic there is cause for considerable concern.

It is well documented that HIV enters the central nervous system (CNS) early after infection and can be associated with neurocognitive impairment in up to 50% of individuals with AIDS (Heaton et al, 1995; Antinori et al, 2007). However, almost all of this data comes from studies conducted in the U.S., Europe and Australia. Little is known about the prevalence and nature of neurobehavioral complications of HIV in developing areas of the world, including China.

In the U.S., research prior to aviability of Highly Active Antiretroviral Treatment (HAART) demonstrated that presence of neuropsychological (NP) impairment in HIV infected persons conferred an increased risk early mortality, above and beyond what could be predicted from medical indices of disease progression (Mayeux et al, 1993; Ellis et al, 1997). It is unclear whether NP impairment still produces worse medical outcomes in the context of HAART, but there is considerable evidence that even mild NP impairment is associated with reduced vocational functioning and other difficulties with cognitive aspects of everyday functioning: medication management, driving, and Instrumental Activities of Daily Living (IADLs; Carter et al, 2003; Heaton et al, 2004a; Hinkin et al, 2004; Marcotte et al, 2006). It is unknown whether any NP deficits that may be observed among HIV infected Chinese farmers would have the same relevance in their lives, or if such deficits could even be noticed by the patients and others around them.

An even more fundamental question is whether Western NP tests will be valid in developing countries, where the people have very different educational, cultural, and linguistic backgrounds from those in the U.S., Europe and Australia. Previous studies in developing countries have been able to use Western NP tests, but estimates of prevalence of HIV-associated neurocognitive disorders (HANDs) from these studies have varied widely. One impediment to reliable estimates of HANDs across studies is the fact that normative data from HIV-uninfected (HIV-) samples matched for age, education, gender, and disease risk factors, were not typically available. This means that some of the “impairment” interpreted as being due to HIV could reflect variation due to other factors. Other reasons for inconsistent findings include use of more or less conservative definitions of NP impairment, differences in the NP tests themselves, and the degree to which investigators have considered comorbid medical and psychiatric conditions that are common in the various populations being studied (e.g., histories of substance use disorders or significant head trauma, other infectious diseases such as malaria, tuberculosis, syphilis and hepatitis C virus - HCV).

So far, the feasibility of cross-cultural NP investigations of HIV effects has been demonstrated best by the WHO Neuropsychiatric AIDS Study Trial (Maj et al, 1994a, 1994b). This study examined the NP status of HIV−, HIV+ asymptomatic and HIV+ symptomatic participants in five countries: Germany, Brazil, Kenya, Zaire and Thailand. A Western NP test battery covering multiple ability domains was translated and successfully administered in the five relevant languages. An increased prevalence of NP impairment was seen among medically symptomatic HIV+ groups (versus their respective HIV− comparison groups) in all sites, whereas asymptomatic participants showed increased rates of impairment in Brazil and Zaire, and in the lowest educated subgroups in Kenya.

More recently, two NP studies of HIV+ groups in sub-Saharan Africa have found very inconsistent results. Robertson et al. (2007) administered the International HIV Dementia Scale (IHDS) and an NP battery covering five ability areas to 110 HIV+ and 100 HIV− adults in Uganda (see also Sacktor et al, 2005; 2006 and Wong et al, 2007 for additional findings with these cohorts). Forty-nine (44.5%) of the infected participants had histories of some ART. Although efforts were made to recruit demographically comparable groups, the HIV− participants were younger (mean 27.5 years versus 36.7 for HIV+), more highly educated (mean 12.1 years, versus 9.1 years for HIV+) and contained fewer females (40% versus 70% for HIV+). Using ANCOVAs with education covaried, the HIV+ group did significantly worse on a mean NP z-score and on all individual tests except for Timed Gait and Grooved Pegboard. Contrasting results were reported by Clifford et al. (2007), who administered the IHDS, and Category Fluency (animals), Timed Gait, Grooved Pegboard and Finger Tapping tests to 73 HIV+ (ART naïve) and 87 HIV− patients in an Ethiopian outpatient clinic. These groups were well matched demographically, and their test performances differed significantly only on Finger Tapping. Clifford and colleagues acknowledged that their test battery was limited, but suggested that their negative results “may be more reliable than studies from other developing countries” (p.70) because their HIV+ and HIV− participants come from the same clinic population, with the same demographic characteristics. Another possibility mentioned by these authors is that the HIV subtype that predominates in Ethiopia (clade C) may have less neurotropism than clade B (predominant in North America, Europe and Australia) and / or clades more predominant in Uganda (clades A & D). However, there is preliminary evidence of NP and electroencephalographic abnormalities being associated with clade C infection in India (Riedel et al, 2006; Sinha & Satishchandra, 2003; Yepthomi et al, 2006).

Finally, Cysique et al. (2007a) recently reported a small study comparing NP effects of HIV infection (clade B) in China and the U.S. An NP test battery that has been used in large multi-site studies in the U.S. was translated and adapted slightly for cultural relevance to the Chinese population, and was administered to 28 HIV+ and 23 demographically similar HIV− controls in China, as well as to 39 HIV+ and 31 HIV− controls in the U.S.. The test battery was well understood and tolerated by the Chinese participants. Importantly, a comparably robust HIV effect on NP performance was found in the two countries. This served as a pilot study for the current investigation of HIV infection among FPDs in Anhui province. There was no overlap in the samples used in the two studies.

All 401 Chinese participants were able to understand and follow the NP test instructions, and obtained what appeared to be valid results. It should be noted that careful translation and piloting of test instruments and the thorough training of the examiners were both important steps in the adaptation and deployment of our NP battery (see Methods & Cysique et al, 2007a for more details). As expected from prior results with Western HIV+ and HIV− cohorts, we found that almost all individual tests in this battery showed some sensitivity to HIV effects in the Chinese FPDs. The single exception was the WCST, which showed no sensitivity to HIV and/or HCV infections in this population. The reason for this is unclear, but it could be that the cultural and/or educational backgrounds of our participants gave them insufficient familiarity with the WCST's stimuli, categorization rules, or the need to repeatedly change rules of responding throughout the test. Regardless of the reason, it appears that this particular test has little or no value in assessing rural Chinese people such as those in the current study.

Clifford et al. (2007) has suggested that some of the reported HIV effect on NP functioning in developing countries may have been biased by the use of HIV− comparison groups that differ from the HIV+ groups in important demographic and background features (e.g., different HIV risk groups). However, the HIV+ and HIV− samples in this study are from the same risk group (FPDs) and have virtually identical demographic and background characteristics (Table 1). Nevertheless, we found robust HIV effects on NP function that are similar to those historically seen in Western countries.

To our knowledge this is the first effort in HIV research within developing countries to use demographically corrected NP norms, based upon results of a large, uninfected group with very similar backgrounds. This permits more accurate classification of acquired HIV-related impairment in individual cases, and arguably provides the best estimates prevalence of HIV-related of impairment. The association of impairment status with biological variables in this study (nadir CD4 and co-infection with HCV) supports the validity of the tests and the norms in this population. On the other hand, generalizability of these norms to populations in other developing countries is uncertain.

The pattern of NP performance in HIV mono-infected individuals is consistent with what has been observed in U.S. cohorts (Heaton et al, 1995), with predominant deficits in processing speed, learning, motor functions and some aspects of executive functions. Also as observed in the U.S., AIDS status was associated with worse overall rates of impairment when compared to non-AIDS participants. The severity of HIV−associated NP impairment observed in this population also is similar to what is seen in Western countries in the era of HAART: almost all (66 of 75) NP-impaired HIV+ participants had GDSs in the mild to moderate range. Only 4% of the total HIV+ group had NP impairment that was severe enough to be consistent with a diagnosis of HIV-associated dementia (Antinori et al, 2007).

Although a majority (56.6%) of our HIV+ FPDs met CDC criteria for an AIDS diagnosis, only 15% had any history of an AIDS defining illness. Most of the AIDS diagnoses occurred because of a prior history of severe immuno-suppression (nadir CD4≤ 200) which then led to their being prescribed modern and effective antiretroviral treatments. This, then, was a relatively healthy HIV+ cohort from the point of view of the proportion of individuals with a history of symptomatic disease. Compared to a large U.S. HIV+ cohort having almost identical rates of AIDS defining illness (Heaton et al, 1995), the current NP methodology (i.e., GDS approach) yielded very similar rates of HIV-associated NP impairment (34.2% in our mono-infected HIV+ group, vs. 31.6% in the U.S. group).

As in studies conducted in Western countries (Letendre et al, 2004; Cysique et al, 2006; Tozzi et al, 2007), we found that in a mixed group of untreated and HAART-treated HIV+ participants, nadir CD4 was associated with NP-impairment, while current CD4 and plasma HIV RNA were not. This cumulative evidence suggests that in the HAART era, brain injury is relatively independent of current immune status and plasma viral load, whereas NP impairment is associated with past histories of more severe immuno-suppression. The fact that we were able to replicate this finding in a very different setting from where these results were initially demonstrated, gives further weight to the importance of historical (versus current) measures of disease status in predicting NP outcomes in the HAART era.

We were able to detect a modest rate of NP impairment among asymptomatic or mildly symptomatic HIV+ participants, whereas the five country study by Maj et al, (1994) found impairment mostly in their symptomatic groups. This difference may be due to differences in ascertainment methods (e.g., different definitions of NP impairment; use of demographically corrected norms). In the U.S. a number of studies have found increased rates of NP impairment in asymptomatic HIV+ groups. In a review of the latter studies, White et al, (1995) reported a median of 31% NP impairment prevalence among medically asymptomatic HIV+ groups; this is consistent with the 29% prevalence rate that we observed in the non-AIDS FPDs in China.

Many published reports from Western cohorts have analyzed NP data in mixed cohorts of treated and untreated volunteers (e.g., Sevigny et al, 2004). This mixture of potentially disparate groups may have accounted for the weakening of the associations between NP impairment and other biological indicators, such as HIV RNA levels, that have been observed in “post-HAART” cohorts. In addition, we found that treated HIV+ individuals in comparison to untreated HIV+ individuals were more likely to be globally impaired (45% vs. 26%, p<.005) and, among those who are impaired, have worse performance (mean GDS 1.1 vs. 0.82, P=.02). The possible explanations for this observation include that treated individuals were also more likely to have AIDS as compared to untreated individuals (75% vs. 30%, p<.001) and that those treated with neurotoxic dideoxynucleoside reverse transcriptase inhibitors (dNRTIs) tended towards having worse global NP performance when compared to those who did not (mean Global Deficit Score 0.61 vs. 0.36; , p=.07).

In addition, the variability of regimens was restricted compared to HIV+ cohorts in Western studies. Only seven different antiretrovirals were reported, with 103 HIV+ participants reporting use of non-nucleoside reverse transcriptase inhibitors (NNRTIs; efavirenz, nevirapine) and 97 reporting use of dideoxynucleoside reverse transcriptase inhibitors (dNRTIs; stavudine and didanosine). Only four individuals reported use of a protease inhibitor (atazanavir (3), nelfinavir (1)). The most common regimen was stavudine-didanosine-nevirapine, reported by 69 HIV+ participants. Lastly, only 16% of treated individuals were on the most neuroeffective regimen (CPE ≥ 2; see Letendre et al., 2008 for additional details on this cut-off) precluding further analysis.

Our screening for HCV infection in this study found a high prevalence of such infection among FPDs. Unlike the situation with HIV infection, however, being HCV+ was not related to the number of reported former plasma donations (12.3 ± 24.1 for uninfected group vs. 16.2 ± 24.5 for HCV mono-infected group; p=.33). The reason for this is unclear. Although we cannot be certain that HCV infection occurred in the same manner as HIV infection (i.e., in the course of plasma donations), two recent surveys in Shanxi province suggest that prior blood donation was a major source of both types of infection (Qian et al, 2005, 2006). Specifically, the latter authors found that: (1) HCV seroprevalence was 27.7% among FPDs in Shanxi, versus only 2.5% in non-FPDs; and (2) co-infection with HCV was seen in the large majority (85%) of HIV infected participants.

We found a small to medium NP effect size among HCV mono-infected individuals in Anhui. This too is in agreement with the current literature in the U.S., which has demonstrated mild neurocognitive impairment in about 30% of HCV+ individuals with mild liver disease, independently of substance use and other co-morbid factors such as depression and fatigue (Forton et al, 2004). Importantly, in our study the combination of AIDS and HCV was associated with the highest rate of NP-impairment (i.e., 50%). This evidence of additive HIV/HCV effects is in accordance with reports in the U.S., which have shown worse neurocognitive deficits in co-infected individuals as compared to both HIV and HCV mono-infection individuals (Ryan et al, 2004; Cherner et al, 2005). However, in the latter U.S. studies, HCV infection was associated with histories of comorbid methamphetamine use disorders, whereas HIV infection was not. The current findings are more clearly interpretable in the absence of drug abuse comorbidity.

In reliably detecting an HCV effect on NP performance, we nevertheless found that the pattern of NP-impairment was different in several ways from HANDs. Specifically, it appeared that individuals with HCV mono-infection were less likely to show deficits on tests of processing speed, attention/working memory and verbal fluency. This should be interpreted with caution, however, given the rather small size of the HCV mono-infected group in the current study (n=51). Additional research with larger HCV mono-infected samples is needed to better establish the prevalence and nature of NP impairments associated with HCV infection among Chinese FPDs. Such research also should examine whether NP impairment is related to active HCV viral replication and/or indices of liver disease, as such findings could have relevance to future clinical care (medical screening, and establishing treatment priorities) within this population.

Our HIV+ cohort in Anhui had lower rates of psychiatric comorbidity than is typically seen in HIV+ groups in Western countries (less current and lifetime substance use disorders and MDD; Atkinson et al, 2008; Bing et al, 2001). Regarding substance use, lifetime alcohol use disorders were not associated with higher rate of NP impairment. Less than 2% of both the overall cohort met criteria for current MDD. Although a somewhat higher prevalence of prior lifetime MDD was seen among HIV mono-infected and co-infected participants (14% vs. 5.0% for controls), this too was lower than what has been reported in U.S. HIV+ cohorts (about 30%) and was unrelated to NP impairment. Similar to what is seen in Western countries, our Chinese HIV+ groups had a modest increase in depressed mood (BDI-II). The size of the association between current depressive symptoms and NP functioning, however, was clinically trivial (shared variance of less than 3%). This is consistent with similar cross-sectional and longitudinal findings from the U.S. (Goggin et al, 1997; Carter et al, 2003; Cysique et al, 2007b), and argues against the possibility that depression occurring in the context of HIV infection is causing the NP impairment. Indeed, the opposite may be the case: some increase in depressed mood may occur in response to awareness of reduced neurocognitive functioning.

As expected, we found associations between NP-impairment and self reports of cognitive complaints as well as decrease independence in IADLs. This finding is in agreement with what has been reported in U.S. cohorts (Heaton et al, 2004a), and demonstrates that NP-impairment has negative effects on daily functioning, even in a non-urban context were most individuals are farmers. The increased cognitive complaints in HIV+ individuals may reflect the fact that these individuals were at least partially aware that some cognitive difficulties could be related to their HIV infection. However, it should be noted that individuals with HCV at the time of testing were not aware of their HCV status. With this in mind, we observed that NP-impaired co-infected individuals reported the worst rates of decline in everyday functioning, as well as more cognitive complaints suggesting an additive effect of the illness.

Lastly, although a repeated finding in the U.S. is that HIV-related NP impairment has been associated with reduced employment (Heaton et al, 1994, 2004a), this was not seen in the current Chinese cohort. Despite the fact that the NP-impaired FPDs tended to be aware of having some cognitive difficulties, they remained active in their work duties. Possible explanations for these different employment outcomes between the U.S. and rural China include: (1) cognitive impairment may have less impact on farming activities in China than on urban jobs in the U.S.; (2) our Chinese participants were mostly self employed, whereas NP impaired people in the U.S. are more likely to be identified by supervisors as having problems; and (3) disability income is more readily available in the U.S., whereas in rural China there is a greater need to keep working.