Factors associated with detection of human immunodeficiency virus in tears: a cross-sectional study

Background Human immunodeficiency virus (HIV) continues to be a global public issue and identifying viral reservoirs is needed in order to better understand the transmission of the disease. We aimed to identify the level of HIV viral load in tears of people living with HIV (PLWH) and study factors influencing their tear viral load.Methods This was a cross-sectional study of patients infected with HIV or acquired immune deficiency syndrome (AIDS) at Beijing You’an Hospital, Beijing, China between May 2018 and July 2018. Sociodemographic information and laboratory test results were collected, including tear samples. Data were analyzed using independent sample t-test, Mann-Whitney U test, Wilcoxon test, Chi-square test, and Spearman correlation analysis.Results 67 subjects met inclusion criteria and were enrolled in the study, of which, 23.9% had detectable HIV ribo nucleic acid (RNA) in their tears. 53.3% of treatment naïve patients and 0.0% of patients who were on anti-retroviral therapy (ART) had detectable HIV RNA in their tears. Tear viral load was lower in tears than in blood plasma and was significantly correlated with plasma viral load (Rho=0.566, p<0.001), but negatively correlated with CD4+T cell count, CD4+/CD8+T cell count, and duration of HIV infection (Rho=-0.450, Rho=-0.464, Rho=-0.565,Rho=-0.252; p<0.001).Conclusions HIV-1 RNA is present in tears, but at a lower detection rate and viral load than in blood plasma. This study provides a potential new avenue for the early diagnosis and linkage to care for patients infected with HIV.

HIV-1 RNA is present in tears, but at a lower detection rate and viral load than in blood plasma. This study provides a potential new avenue for the early diagnosis and linkage to care for patients infected with HIV.

Background
It is widely recognized that human immunodeficiency viruses (HIV) can penetrate into various tissues and exist in many bodily fluids and secretions, such as blood, seminal plasma, cerebrospinal fluid, breast milk, and saliva (1-2). Fortunately, highly active antiretroviral therapy (HAART) cannot only protect the health of infected people through suppression of HIV replication, but also improve the overall health and wellbeing of those infected [3] . However, more and more studies have discovered that the relationship between HIV viral load in blood and bodily fluids is not consistent, as many factors contribute to the detection of HIV ribonucleic acid (RNA) in patients regardless of HAART status (4-7). Thus, further investigations into this phenomenon demand our attention.
Tears play a crucial role in refraction, preventing infection, and maintaining homeostasis of ocular surfaces (8). Studies have demonstrated that HIV-positive patients are more likely to suffer from anterior segment diseases such as dry eye (9)(10)(11), compared to the general population without HIV infection. While previous studies have found that it was hard to isolate HIV-1 and detect pro-viral sequences in tears of HIV-positive patients (12), recent research has suggested that patients who underwent long-term HAART with undetectable plasma viral load had detectable HIV-1 viral load in their tears (13). This suggests that the lacrimal gland could be a new reservoir for the HIV virus.
However, few available data have reported on the dynamics of HIV in tears. Therefore, our objectives were to measure the level of viral load in tears of people living with HIV (PLWH) and study factors influencing their tear viral load.

Study Participants
All patients infected with HIV or acquired immune deficiency syndrome (AIDS) were randomly recruited from the Infectious Diseases Center or Ophthalmology Department of the Beijing You'an Hospital, Beijing, China between May 2018 and July 2018. Eligible participants included all patients with confirmed HIV infection, diagnosed by infectious disease specialists at You'an Hospital. Patients whom were too sick to complete an eye examination or refused to participate were excluded from the study. After screening for eligibility, the study team obtained written and oral consent. Patients meeting inclusion criteria completed an in-person questionnaire delivered by a doctor trained in study ethics.

Sociodemographic characteristics
Information including demographic characteristics (age, sex, race, smoking status, alcohol use), likely mode of HIV transmission (homosexual, heterosexual, injection drug use), date of HIV diagnosis, medical history, co-infections and opportunistic infections, type of HAART regimen used, and duration of HAART treatment, were collected from all participants.
Additionally, participants' laboratory test results, including peripheral blood lymphocyte count and presence of a sexually transmitted infection (STI) were collected. STIs were tested using treponema pallidum particle agglutination assay (TPPA) and rapid plasma reagin circle card assay (RPR). Lastly, plasma HIV-1 viral load was extracted from medical records.

Tear collection
A well-trained ophthalmologist collected tears from patients included in the study in an ultravioletdisinfected room. 0.2-1.5 milliliters (mL) of tears were collected from the conjunctiva sac directly and stored at−80°C refrigerators at the hospital.

HIV RNA quantification
The primary outcome of interest was presence and quantification of HIV RNA in tears. Qualitative and quantitative collection of HIV-1 RNA in tears and blood plasma was performed according to manufacturer's instructions using Abbott M2000 Real Time HIV-1 assay (Abbott Molecular, Inc.). HIV RNA results were stratified as TND (target not detected), below 40 copies/mL, or above40 copies/mL.

Statistical analysis
Data collected by the questionnaire and blood plasma and tear testing were directly imported into . Kolmogorov-Smirnov test was used for normality testing. Normal variables were presented as the mean and standard deviation (SD) and non-normal variables as the median and interquartile range (IQR). T-tests were used to compare the means of normally distributed quantitative variables, otherwise, a Mann-Whitney U test was used. The nonparametric Wilcoxon test was used to compare tear and plasma viral loads. Chi-square tests were applied to compare qualitative data, such as the detection rate of HIV RNA in tears and blood, and in treatment-naïve patients and HAART-treated patients. Spearman's rho test was computed for measuring associations of tear viral load. The values of blood plasma and tear viral load were transformed to log10 prior to logistic regression analysis. A value of p < 0.05 was considered as statistically significant.

Characteristics of study population
67 patients met inclusion criteria and were included in the study, among which 30 were treatmentnaïve patients and 37 were HAART-treated patients; 65 were male; and the mean age was 35±9 years (Table 1). Of all participants, 73.1% were infected due to homosexual contact.  Furthermore, of all participants, tear viral load ranged from TND to 13,096 copies/mL and median viral load was generally lower in tears than in blood plasma (0 vs. 156 copies/mL, p < 0.001), as seen in Table 2. Additionally, patients not yet initiating HAART-treatment were also found to have a lower median viral load in tears compared to in blood plasma (40 vs. 22,555.5 copies/mL, p < 0.001)

Factors associated with tear viral load in all participants
As seen in Table 3 and Figure 1, viral load in tears and in blood plasma were found to be positively

Factors associated with tear viral load in treatment-naïve participants
Of 30 treatment-naïve subjects, tear viral load was only associated with duration of infection   The scatter diagram summarizes the distribution of patients by viral load status and antiretroviral therapy status. Among 67 subjects, viral load was generally lower in tears than in blood plasma.
Among 30 treatment-naïve patients, more than half had detectable HIV RNA in tears, while among 37 patients who were on antiretroviral therapy, 0.0% had detectable HIV RNA in tears.

Discussion
With a better understanding of HIV persistence in various bodily fluids, ophthalmologists are becoming especially interested in the existence of HIV RNA in tears and factors associated with tear viral load. However, few studies have focused on this issue. Our study aimed to gather baseline information regarding the existence of HIV RNA in tears and associated factors, which may lead to presence of viral load in tears. The results of this study can be used to discuss potential HIV reservoirs and inform the integration of a holistic approach to HIV care, not just amongst infectious disease specialists but patients' spectrum of providers.
In this cross-sectional study, 23.9% of participants had detectable HIV RNA in their tears. The presence of tear viral load was also associated with blood viral load, CD4+ T cell count, CD4+/CD8+ T cell count and duration of HIV infection. Furthermore, the detection rate and viral load were lower in tears than in blood and lower in HAART-treated patients than in untreated patients. This is similar to other studies among varying bodily fluids and secretions, such as cerebrospinal fluid, saliva, breast milk, cervicovaginal lavage fluid and semen (14)(15)(16)(17)(18)(19) HIV may survive and duplicate in eyes without severe restriction (20). Previous studies have shown that only a small subset of HIV variants could intrude into the eye tissue (21). Therefore, it is of great importance to keep exploring the possibility of HIV reservoirs (22) in eyes and factors associated with the presence of HIV in tears. To identify factors associated with tear viral load in treatment-naïve subjects, we performed various correlation and regression analyses. However, no variables were found to significantly impact viral load except the duration of infection. This lack of association may be the result of the relatively small sample size or the existence of outlier data (Fig. 1). Indeed, when we exclude these four extremes (blood viral load >150000copies/ mL or tear viral load >10000copies/ mL), there is a significant relationship between tear viral load and blood viral load (Rho=0.43, p=0.03). This phenomenon was affirmed when we retested the outlier samples, and it was also reported in the study of other bodily fluids (e.g. saliva, cerebrospinal fluid, aqueous humor) (23)(24)(25)(26)(27).
Considering this is the first study exploring tear viral load in treatment-naïve HIV/AIDS, we still lack sufficient evidence to explain our findings now. Future studies of larger study cohorts are needed to verify our results and establish a more robust association, meanwhile, more histopathology and immunohistochemistry examinations are required to point out the possible mechanism of HIV infection in eyes.

Conclusions
According to the results of this study, HIV-1 RNA is present in tears, but at a lower detection rate and viral load than in blood plasma. Tear viral load is positively associated with plasma viral load while it is negatively correlated with CD4 cell count, with HAART treatment reducing viral load in tears. This study provides more insight into the HIV viral dynamics in eyes and opens up new avenues of early diagnosis, intervention, and prognosis of the disease.
Abbreviations AIDS: Acquired immune deficiency syndrome reasonable request.

Competing interests
The authors declare that they have no competing interests