HIV Type 1 and Cytomegalovirus Coinfection in the Female Genital Tract

Abstract

The relationship between human immunodeficiency virus(HIV) type 1 and human cytomegalovirus (CMV) was studied in blood, saliva, and cervicovaginal lavage (CVL) specimens from 33 HIV-1–infected women. An association between HIV-1 RNA and CMV DNA was found in the CVL specimens, which also were tested for cytokine levels. Women with detectable CMV DNA in CVL specimens were more likely to have higher interleukin (IL)–1β and IL-8 levels than were women with undetectable CMV DNA in CVL specimens. More than 1 strain of CMV was detected in specimens from 2 patients. These results suggest mechanisms by which CMV coinfection could affect HIV-1 disease progression.

Shedding of HIV type 1 in semen is a common route of transmission between HIV-1–infected men and their sexual partners. Several studies have suggested that cytomegalovirus (CMV) coinfection may enhance HIV-1 shedding and increase the risk of progression to AIDS [1–3]. The presence of multiple CMV strains in seminal fluid from individual patients also has been associated with an increased risk of AIDS [4].

There are fewer studies of HIV-1 and CMV coinfection in women. CMV shedding in cervicovaginal secretions from HIV-1–infected women has been reported [5–7]. However, no studies of women have directly examined the association of CMV DNA with HIV-1 RNA or coinfection with multiple CMV strains in the genital tract.

CMV infection is associated with increased production of several cytokines, including tumor necrosis factor (TNF)–α, interleukin (IL)–6, and IL-8 [8, 9]. IL-8 has been reported to enhance HIV-1 replication [10]. Thus, up-regulation of cytokines is a potential mechanism for CMV interaction with HIV-1 in coinfected compartments.

We report the results of a study of a cohort of HIV-1–infected women. The association of CMV DNA and HIV-1 RNA was determined in blood, cervicovaginal lavage (CVL), and saliva specimens. Cytokine levels in blood and CVL specimens also were determined. DNA sequencing analysis was performed on selected specimens, to detect simultaneous infection by multiple strains of CMV.

Patients and methods

Study patients were enrolled in the Division of AIDS Treatment Research Initiative (DATRI) 009 substudy of the Women’s Interagency HIV Study (WIHS) [11] and Women’s Health Study 001 [12]. Specimens from 55 HIV-1–infected women who were asymptomatic for CMV disease were tested for CMV DNA. A cohort of 33 women was selected for further study on the basis of availability of CVL specimens containing cells. Informed consent was obtained from the patients participating in the DATRI 009 substudy of the WIHS. The human experimentation guidelines of the US Department of Health and Human Services, Rush University Medical Center, and the University of Southern California Medical Center were followed in the conduct of this research.

Testing for CMV DNA was performed on 150 specimens from the 55 HIV-1–infected women, including (1) CVL specimens either unfractionated or fractionated into cells and supernatant, (2) saliva specimens, and (3) plasma and white blood cell specimens. Three specimens from 2 different women contained nonviable CMV from infected cells in culture: 2 specimens were obtained from the genital tract, and 1 was from saliva.

HIV-1 RNA in all specimens had been assessed previously using the Roche Amplicor 1.0 assay (Roche Molecular Systems). The results were quantitative but were converted to positive or negative on the basis of the lower limit of detection (400 copies/mL), for the purpose of comparison with the qualitative CMV DNA assay described below.

The lysis and extraction reagents from the NucliSens pp67 CMV assay (bioMérieux) were used to extract CMV DNA, which was detected using polymerase chain reaction (PCR) amplification with primers 5′-CTGCACAACGTCACGGTACATC-3′ (forward) and 5′-CGGAAAGCAGGGTGGTAACATTC-3′ (reverse). The 368-bp product from a highly conserved region of the UL97 gene [13] was confirmed by Southern blot hybridization using an internal probe (284 bp) produced by PCR amplification with primers 5′-GAAGCTGGCGTGCATCGACA-3′ (forward) and 5′-TAACATTCGCGCAGACGGTG-3′ (reverse). Negative PCR amplifications were reamplified by nested PCR using primers 5′-CAGTGGAAGCTGGCGTGCAT-3′ (forward) and 5′-GGTAACATTCGCGCAGACGGTG-3′ (reverse).

A portion of the hypervariable UL144 open reading frame (ORF) [14] was amplified from cultures and specimens from 2 women, by use of primers 5′-TCGTATTACAAACCGCGGAGAGGAT-3′ (forward) and 5′-ACTCAGACACGGTTCCGTAA-3′ (reverse). The PCR products were sequenced using the ABI Prism BigDye Terminator Cycle Sequencing kit (Applied Biosystems) and were analyzed on an ABI 3100 Automated DNA Sequencer (Applied Biosystems). A mixture of sequences was indicative of >1 CMV strain. PCR products containing mixtures of sequences were cloned using the pGem-T Vector System (Promega). Plasmids extracted from individual cloned colonies were sequenced to determine the number of CMV strains present.

Cytokines in plasma and CVL supernatant specimens were measured quantitatively using commercial kits for human interferon (IFN)–γ, human IL-1β, human IL-6, human IL-8, and human IL-10 (Endogen, Pierce) and for TNF receptor 2 (TNFR2) (sTNF-RII, BioSourceEurope). The sensitivities of the assays are as follows: IFN-γ, <2 pg/mL; IL-1β, <1 pg/mL; IL-6, <1 pg/mL; IL-8, <2 pg/mL; IL-10, <3 pg/mL; and TNFR2, <0.1 ng/mL.

All statistical tests were performed using SAS software, version 8.2 (SAS Institute). Univariate analyses were conducted to compare the characteristics of subjects who were positive for CMV DNA with those who were negative, as determined by testing of specimens from the genital, blood, and oral compartments. The χ² test was used for categorical variables. Fisher’s exact test was used for variables with small cell frequencies. Because of the wide variation in HIV-1 RNA levels, the Kruskal-Wallis procedure was used to test the median difference between CMV DNA detection and HIV-1 RNA level in the 3 compartments (genital, blood, and oral). A t test was performed to assess the association between mean cytokine values and CMV DNA positivity.

To determine the relationship between CMV DNA detection and HIV-1 RNA level and the association between CMV DNA in CVL specimens and antiretroviral therapy (ART), a series of logistic regression models was used with and without adjustment for potential risk factors. The covariates considered were absolute CD4⁺ cell count (<200, 200–499, and ≥500 cells/mm³) and ART. For women who had undetectable HIV RNA and/or cytokine levels, half the lower limit of detection was used for all analyses, unless indicated otherwise. All statistical tests were based on 2-tailed alternatives with an α level of 0.05.

Results

We originally analyzed blood, saliva, and CVL specimens from 55 HIV-1–infected women. The most consistent results were obtained from CVL specimens containing cells (unfractionated or cell pellets); however, these specimens were not available from 22 of the 55 women. These 22 women were not included in further analyses of the CVL specimens, although 3 of them had detectable CMV DNA in CVL supernatant specimens. The patient cohort consisted of 33 women (table 1) from whom CVL cellular specimens were available. Exclusion of the 22 women did not change the demographic and clinical characteristics of the study cohort (n = 33), which are summarized in table 1.

Table 1.

Demographic and clinical characteristics of the study cohort.

Characteristic	No. (%) of patients (n = 33)a	Mean (SD)	Median
Demographic
Age, years		34.09 (6.67)	34.00
Race
White	5 (15.15)
Black	22 (66.67)
Other	6 (18.18)
Antiretroviral therapy
None	12 (37.50)
Monotherapy	2 (6.25)
Combination without PI	6 (18.75)
Combination with PI	12 (37.50)
Clinical covariate
CD4+ cell count, cells/mm3		400.64 (263.23)	359.00
<200	6 (18.18)
200–499	18 (54.55)
≥500	9 (27.27)
HIV RNA level, copies/mLb
Blood		163,822 (638,584)	15,000
Detectable	26 (78.79)
Undetectable	7 (21.21)
CVL		5976 (12,812)	80
Detectable	17 (51.52)
Undetectable	16 (48.48)
Saliva		9866 (21,964)	2200
Detectable	7 (58.33)
Undetectable	5 (41.67)
Cytokine level, pg/mL
TNFR2
CVL supernatant		137.93 (166.13)	50.00
Plasma		6154.38 (4558.71)	4746.00
IFN-γ
CVL supernatant		2.33 (5.87)	1.00
Plasma		1.00 (0.00)	1.00
IL-1β
CVL supernatant		51.62 (124.61)	7.47
Plasma		0.98 (1.12)	0.50
IL-6
CVL supernatant		17.31 (42.86)	5.90
Plasma		0.84 (0.74)	0.50
IL-8
CVL supernatant		913.69 (1718.87)	179.19
Plasma		3.03 (4.68)	1.00
IL-10
CVL supernatant		19.93 (83.64)	1.50
Plasma		1.81 (0.83)	1.50

NOTE.

Combination antiretroviral therapy was defined as 2 nucleoside reverse-transcriptase inhibitors. CVL, cervicovaginal lavage; IFN, interferon; IL, interleukin; PI, protease inhibitor; TNFR2, tumor necrosis factor receptor 2.

^aBecause of missing data, the no. of patients in each category may not sum to the total sample size.

^bHalf the lower limit of detection was used to estimate values for women with undetectable HIV RNA measurements.

Approximately 75% of women with detectable CMV DNA in CVL specimens had detectable HIV-1 RNA in CVL specimens (table 2). For the women who were CMV DNA positive, the median HIV-1 load was significantly higher than that for the women who were CMV DNA negative (P =.005). Results from the logistic regression models indicated that, when other cofactors were not considered, the odds of being CMV DNA positive were ~5 times greater for women with detectable HIV-1 RNA in CVL specimens than for those with undetectable HIV-1 RNA (odds ratio [OR], 4.88; 95% confidence interval [CI], 1.01–23.57). The association became statistically even stronger after adjustment for absolute CD4⁺ cell count (OR, 6.24; 95% CI, 1.02–38.12).

Table 2.

Factors associated with detection of cytomegalovirus (CMV) DNA in cervicovaginal lavage (CVL) specimens from 33 HIV-1–infected women.

	Detection of CMV DNA
				Unadjusted		Adjustedc
Variable	Positivea	Negativea	Pb	OR (95% CI)	P	OR (95% CI)	P
HIV RNA level, copies/mL	4487	63	.005
Detectable	9 (75.00)	8 (38.10)	.041	4.88 (1.01–23.57)	.049	6.24 (1.02–38.12)	.048
Undetectable	3 (25.00)	13 (61.90)		1.00		1.00
Receiving antiretroviral therapyd
Yes	6 (54.55)	14 (66.67)	.501	0.6 (0.14–2.67)	.503	0.43 (0.08–2.25)	.319
No	5 (45.45)	7 (33.33)		1.00		1.00
CD4+ cell count, cells/mm3
<200	3 (25.00)	3 (14.29)	.520	1.25 (0.16–9.92)
200–499	5 (41.67)	13 (61.90)		0.48 (0.09–2.56)	.833
≥500	4 (33.33)	5 (23.81)		1.00	.390

NOTE.

CI, confidence interval; OR, odds ratio.

^aData are no. (%) of patients (n = 33) for the categorical variables. Median values are given for the continuous variable HIV RNA level.

^bDetermined by the Kruskal-Wallis test, for the continuous variable, or by the χ² or Fisher’s exact test, for the categorical variables.

^cThe model was adjusted for CD4⁺ cell count (<200, 200–499, and ≥500 cells/mm³).

^dBecause of missing data, the no. of patients in each category does not sum to the total sample size.

CMV DNA was not detected in any of the blood specimens (plasma or cells). Saliva specimens were available from only 12 of the 33 women. Four of these women were coinfected, 2 had detectable CMV DNA only, 3 had detectable HIV-1 RNA only, and 3 were uninfected. Thus, these numbers are too small to draw any conclusions.

The relationship between CMV DNA detection and cytokine production in blood and CVL specimens was examined as a possible mechanism for CMV and HIV-1 interaction. The mean and median values for the cytokines in the CVL and plasma specimens are shown in table 1. The availability of data for cytokine production from only 29 of the 33 women provided little power for the detection of statistically significant relationships. However, an assessment of patients with detectable cytokine levels showed that, for women with detectable CMV DNA in CVL specimens (n = 11), the mean IL-8 level in CVL specimens was 1941 pg/mL, compared with a mean IL-8 level of 285 pg/mL for women with undetectable CMV DNA (n = 18) (P = .053). The mean IL-1β level in CVL specimens from women with detectable CMV DNA (n = 11) was 181 pg/mL, compared with a mean of 18 pg/mL for women with undetectable CMV DNA (n = 18) (P = .09). TNFR2, IFN-γ, IL-6, and IL-10 did not show comparable trends for higher levels in women with detectable CMV DNA in CVL specimens, compared with those with undetectable CMV DNA. HIV-1 RNA shedding in CVL specimens also was moderately associated with IL-1β level (r = 0.41; P = .0285) and IL-8 level (r = 0.43; P = .0191) in CVL specimens.

Infected-cell cultures that no longer contained viable CMV were available for 2 women. This material and the specimens (CVL and saliva) from these 2 patients yielded sufficient DNA for sequencing analysis. We compared the CMV strain sequences within and among compartments, as represented by all the specimens obtained from these 2 women. For the remaining women, additional sequencing analysis was not done, because sufficient specimen material was not available.

A rapid screen for the presence of multiple CMV strains was developed on the basis of the hypervariable UL144 ORF [14]. A mixture of sequences in the hypervariable region indicated the presence of >1 CMV strain in specimens from both women. Subcloning and sequencing of the PCR products demonstrated differences among the strains when detected in culture, compared with direct analysis of the specimens from different compartments. In the culture of CVL fluid from the first patient, only a UL144 group 1A strain was detected, whereas UL144 group 1A and 3 strains were detected in the corresponding CVL specimen. In the culture of CVL fluid from the second patient, only a UL144 group 1B strain was detected, whereas group 1A, 1B, and 2 strains were detected in the CVL specimen. The saliva culture from the same patient contained mixtures of group 1A and 1B strains. These results suggest that culture does not always detect all infecting strains and that individual compartments may be infected with different strains.

Discussion

CMV detection and genotypic data for the women in this study parallel results from studies of men in whom an association between HIV-1 RNA and CMV DNA in the genital tract was found. Although a previous study found an association between cervical HIV-1 DNA and CMV DNA [7], to our knowledge the present study is the first to assess CMV DNA and HIV-1 RNA in the female genital tract and to detect multiple CMV strains within this compartment during HIV-1 coinfection.

Women who had detectable CMV DNA in CVL specimens were more likely to have detectable HIV-1 RNA in CVL specimens. HIV-1 RNA, but not CMV DNA, was detected in the blood of women whose CVL specimens were positive for both CMV DNA and HIV-1 RNA. The median HIV-1 RNA level in CVL specimens from women with detectable CMV DNA in CVL specimens was higher than that in women with undetectable CMV DNA. Thus, coinfection may lead to CMV and HIV-1 interaction in the female genital tract. CD4⁺ cell counts and ART regimens did not appear to be associated with detection of CMV DNA in CVL specimens (table 2), suggesting that the genital compartment may serve as a sanctuary for these viruses. However, some of the ART regimens used at the time of the DATRI 009 substudy were suboptimal; thus, current regimens might have different results.

CMV could enhance HIV-1 expression either in coinfected cells or in singly infected cells in close proximity in the same compartment, and cytokines could serve as mediators of this interaction. Our findings of up-regulation of HIV-1 shedding are similar to those reported by others for HIV-1 infection alone, in which elevated cytokine levels in the vaginal compartment correlated with vaginal HIV-1 shedding but not with plasma HIV-1 load [15]. IL-8 is of particular interest, because up-regulation of IL-8 has been reported during both CMV and HIV-1 infection; in turn, HIV-1 replication can be enhanced by IL-8 [8–10]. Our results show a trend in which the mean IL-8 level in CVL specimens was higher among women whose CVL specimens contained detectable CMV DNA. This leads us to postulate that CMV induces a proinflammatory state in the genital tract that activates HIV-1 and that ART does not have an impact on this interaction.

Because this was a cross-sectional, retrospective assessment of a limited number of women, our ability to define the interaction of CMV and HIV-1 in the genital tract and the subsequent clinical outcome also was limited. Our results indicate that a longitudinal study of CMV and HIV-1 interaction in the female genital tract would provide a better understanding of the relative variability of CMV and HIV-1 loads over time, the impact of ART, and the effect of multiple infecting strains on disease progression.