Baseline Levels of Soluble CD14 and CD16⁺56⁻ Natural Killer Cells Are Negatively Associated With Response to Interferon/Ribavirin Therapy During HCV-HIV-1 Coinfection

Abstract

Disease progression of human immunodeficiency virus type 1 (HIV-1) is associated with immune activation. Activation indices are higher during coinfection of hepatitis C virus (HCV) and HIV. The effect of immune activation on interferon α (IFN-α) therapy response is unknown. We evaluated soluble CD14 (sCD14) and natural killer (NK)–cell subsets at baseline, and during pegIFN-α2a/ribavirin therapy in HCV-HIV coinfection. The sCD14 level increased during therapy. Baseline sCD14 positively correlated with baseline HCV level and CD16⁺56⁻ NK-cell frequency, and both sCD14 and CD16⁺56⁻ NK cells correlated negatively with magnitude of HCV decline. IL28B genotype was associated with therapy response but not sCD14 or CD16⁺56⁻ NK frequency. Markers of innate immune activation predict poor host response to IFN-α–based HCV therapy during HCV-HIV coinfection.

Persistent activation of the immune system is well described during progressive human immunodeficiency virus type 1 (HIV-1) infection. This can be quantified by soluble markers including cytokines and sCD14, as well as markers of T, B, natural killer (NK), and myeloid cell activation [1–3]. Downstream effects of immune activation include increased T-cell turnover and increased availability of targets for HIV infection [2]. Effects such as these may directly contribute to impaired host control of HIV and other pathogens [4]. Hepatitis C virus (HCV) infection is also associated with elevated soluble CD 14 (sCD14) levels and activated CD8 T cells [5, 6], and these markers may likewise predict mortality and disease stage [5, 6].

Factors associated with response to pegylated interferon-α (pegIFN)/ribavirin (RBV) therapy for HCV infection include HCV genotype, age, race, HIV coinfection, HCV-plasma level, and IL28B gene locus polymorphism [7–9]. Mechanisms underlying these associations are largely unknown, although the integrity of the host immune system likely contributes [4, 10]. Whether immune activation attributable to HIV or HCV infection plays a role in determining or predicting outcome of HCV therapy is not known. On the basis of previously observed associations between NK-cell subset frequency and phenotype, interferon α (IFN-α)–based therapy outcome, and sCD14 level during HCV monoinfection [4, 5, 10, 11], we evaluated the relationship between sCD14, NK-cell subset frequencies, and outcome of IFN-α based therapy for HCV in the setting of HIV-1 coinfection.

METHODS

AIDS Clinical Trials Group (ACTG) trial A5071 was a prospective randomized trial evaluating effectiveness and safety of pegIFN-α2a/RBV in comparison to nonpegylated-IFN/RBV in the setting of HCV-HIV coinfection [8]. Entry criteria included at least 12 weeks on antiretroviral therapy (ART) with CD4 cell count >100/mm³ and HIV-1 RNA < 10 000 copies/mL, or CD4 cell count >300/mm³ if ART naive. In the study arm given pegIFN/RBV 7 of 51 subjects with HCV genotype-1 infection achieved a sustained virologic response, as compared to 11 of 15 of subjects with other genotypes [8]. After obtaining ACTG site IRB approved consent, peripheral blood mononuclear cell (PBMC) and plasma samples were prepared at each clinical site by standardized ACTG protocol, cryopreserved, and then sent to a central storage facility. We used plasma and cryopreserved PBMCs available from baseline, and plasma from baseline and weeks 12 and 24 of pegIFN-α/RBV arm subjects with week-24 viral load available. Samples were available for 28 subjects (22 genotype-1 HCV). A subset of these subjects (n = 18) also consented to use of DNA and were included in the IL28B genotype analysis.

Plasma was evaluated for sCD14 level by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, Minnesota).

Cryopreserved PBMCs were thawed and stained with Violet Live/Dead stain (Invitrogen, Grand Island, New York), anti-CD3-PerCP (clone SK7), anti-CD14-APC (clone M5E2), anti-CD19-APC (clone SJ25C1), anti-CD16-APC-Cy7 (clone 3G8), anti-CD56⁻PE-Cy7 (clone NCAM16.2), anti-NKp30-PE (clone P30-15) (BD Biosciences, San Jose, California), and anti-IFNαR1-FITC (clone 85228) (R&D Systems), or isotype controls. Flow cytometry data were acquired on a BD LSRII flow cytometer (BD Biosciences), and analyzed using FACS Diva Software (BD Biosciences). Live cells were identified by forward and side scatter and viability stain (median proportion of lymphocyte gated cells that were viable, 95.6%; interquartile range [IQR], 90%–97%). We quantified proportions of viable lymphocyte gated cells that were CD3-CD14-CD19-CD16⁺CD56⁺, CD3-CD14-CD19-CD16-CD56^+/hi, CD3-CD14-CD19-CD16⁺CD56⁻. NKp30 and IFNαR expression were recorded as mean fluorescence intensity above isotype control (Supplemental Figure 1).

IL-28B single-nucleotide polymorphisms (SNPs) rs8099917 and rs12979860 were genotyped using the ABI Taqman Allelic Discrimination Assay (Applied Biosystems, Carlsbad, California). Primers and probes for rs12979860 were as described elsewhere [12]. For rs8099917, forward primer GCCTGTCGTGTACTGAACCA, reverse primer GCGCGGAGTGCAATTCAAC, probe (C allele VIC-TGGTTCGCGCCTTC, T allele FAM-CTGGTTCACGCCTTC designed by Applied Biosystems). Thermal cycling was performed on an ABI-7500-HT Sequence Detector system. Fluorescence data were collected and genotypes determined using SDS software (Applied Biosystems).

Statistical analyses were performed using SPSS for Windows v. 20.0 (IBM Corp, Armonk, New York). Associations between continuous variables were evaluated using Spearman rank correlation coefficient. Group comparisons were analyzed by Mann-Whitney U test or the Kruskal-Wallis test, as appropriate. Virologic responses were defined as follows: rapid virologic response (RVR) is undetectable HCV in plasma (<600 IU/mL) at week 4; early virologic response (EVR) is at least a 2-log-10 HCV decline at 12 weeks; complete EVR (cEVR) is undetectable HCV at 12 weeks of therapy; and sustained virologic response (SVR) is undetectable HCV 6 months after completion of HCV therapy. The Jonckheere-Terpstra test for continuous variables, and the Cochran-Armitage test for proportions, were used to analyze trends across the 3 IL28B genotypes. All tests of significance were two-sided and P values ≤ .05 were considered significant.

RESULTS

Baseline characteristics of the 28 study subjects are shown in Supplemental Table 1. Overall, median age was 42 years, 75% of subjects were men, and 53.6% were nonwhite. In total, 78.6% had genotype-1 HCV infection, and 14% were ART naive.

We measured sCD14 at baseline and over the course of pegIFN/RBV therapy. As shown in Figure 1A, sCD14 levels increased over the first 12 weeks of therapy. We next addressed our primary objective, evaluating whether there is a relationship between sCD14 and viral decline during pegIFN/RBV therapy in the setting of HCV-HIV coinfection. Because the rate of viral decline varies by HCV genotype, we restricted the analysis to genotype-1 infected subjects (n = 22). We observed a negative correlation between baseline plasma sCD14 and magnitude of HCV decline that was significant at week 4 (r = −0.87, P < .001) and trending toward significance at week 12 (r = −0.46, P = .09) and week 24 (r = −0.51, P = .08) (Figure 1B). Additionally, baseline HCV level positively correlated with sCD14 (r = 0.552 P = .04). However, baseline HCV level was not correlated with HCV decline magnitude.

Figure 1.

Plasma soluble CD14 (sCD14) increases during pegylated interferon-α/ribavirin (pegIFN/RBV) killer (NK)–cell frequency. A, Plasma sCD14 level was measured at baseline, week 12, and week 24 of pegIFN/RBV therapy (n = 19, all HCV genotypes). B, C, Baseline plasma sCD14 level (B) and CD3–14-19-16⁺56⁻ NK-cell frequency (C)in relation to week 4 (n = 14 for sCD14 and n = 19 for CD16⁺CD56⁻ NK), week 12 (n = 14 for sCD14, and n = 20 for CD16⁺CD56⁻ NK), and week 24 (n = 13 for sCD14 and n = 19 for CD16⁺CD56⁻ NK) HCV genotype-1 viral decline magnitude.

We also analyzed baseline, week 12, and week 24 sCD14 levels in relation to therapy response status (RVR vs non-RVR, EVR vs non-EVR, cEVR vs non-cEVR, and SVR vs non-SVR). Baseline sCD14 was associated with SVR status (P = .01, Figure 2). Week-12 sCD14 was also associated with viral response status (cEVR, P = .04; SVR, P = .04), whereas week-24 sCD14, change in sCD14 between baseline and week 12, week 24, or change in sCD14 between weeks 12 and 24 did not correlate with viral response status.

Figure 2.

Pairwise comparisons for hepatitis C virus (HCV) genotype 1 virologic response status. A, Plasma sCD14 at baseline (week 0) and week 12 is plotted by virologic response status (rapid virologic response [RVR], early virologic response [EVR], complete EVR [cEVR], and sustained virologic response [SVR]). B, Week 0 CD16⁺56⁻ natural killer (NK) frequency by virologic response status. Box plots represent median, interquartile range, minimum, and maximum. P < .1 is shown.

In agreement with our earlier finding, CD16⁺CD56⁻ NK-cell subset frequency at baseline was found to negatively correlate with HCV genotype-1 viral decline magnitude at week 4 (r = −0.45, P = .05) and week 24 (r = −0.55, P = .01), and trending toward significance at week 12 (r = −0.39, P = .09) (Figure 1C). These findings held after removing a subject with an extremely high frequency of this cell population (data not shown). Furthermore, we observed baseline sCD14 to correlate with CD16⁺CD56⁻ NK-cell subset frequency (r = 0.49, P = .045). In contrast to sCD14, CD16⁺CD56⁻ NK cells did not correlate with baseline HCV load level. Subjects who achieved virologic response had lower baseline CD16⁺CD56⁻ NK-cell subset frequency than those who did not (EVR P = .03, SVR P = .02, Figure 2). In contrast to our prior finding in the setting of HCV monoinfection [11], we did not find a correlation between CD16⁺CD56⁻ NK-cell IFN-αR expression at baseline and magnitude of HCV decline in HCV-HIV coinfection, although CD16⁺56⁻ NK-cell IFN-αR expression did tend to be higher on cells of subjects with an RVR than those without, when subjects with all HCV genotypes were analyzed (P = .05).

IL28B genotype is associated with magnitude of HCV decline in the setting of IFN-α–based therapy for HCV infection [7]. We were able to perform IL28B rs12979860 and rs8099917 genotype analysis on samples from 18 subjects in this data set. IL28B rs12979860 genotype distribution was n = 2 TT, n = 11 CT, n = 5 CC, and for rs8099917 genotype distribution was n = 14 TT, n = 4 GT, n = 0 GG. Race did not differ significantly among IL28B genotypes (P = .78). As expected, IL28B rs1297860 genotype was associated with a nonsignificant trend in 1-month viral decline magnitude among genotype-1 infected subjects (median log₁₀ decline 0.25, 0.45, and 2.10 in TT, CT, and CC groups, respectively, P for trend = .1; and the corresponding proportions of genotype-1 infected subjects reaching an RVR were 0, 0 and 50%, P for trend = .048). Conversely, rs8099917 was not significantly associated with HCV decline or response. In addition, IL28B genotype did not associate with sCD14 or NK-cell subset frequency. Likewise, sCD14 levels and NK-cell subset frequency did not show consistent differences by sex or race.

DISCUSSION

Soluble CD14 is released from monocytes upon activation [13], and elevated plasma levels have been associated with mortality during HIV-1 infection [5] and associated with disease stage and treatment outcome in the setting of HCV monoinfection [5]. Soluble CD14 was observed here to increase over the early course of IFN-α–based therapy for HCV in HIV-coinfected patients. This increase has, to our knowledge, not been described before in the setting of HCV or HIV infection. Because IFN-α is a broad immune activating cytokine, one possible interpretation is that IFN-α directly, or indirectly, contributes to monocyte activation and sCD14 release. In regard to the latter, it is interesting to note that plasma IP10, a well-known interferon stimulated chemokine associated with therapy response [14], tended to correlate here with sCD14 level (P = .1) and did correlate with CD16⁺56⁻ NK-cell frequency (P = .02; not shown).

Baseline and week-12 plasma sCD14 levels were found to negatively associate with the magnitude of HCV load decline. Thus, similar to the previously described negative predictive value of higher baseline plasma IP10 level [14], this marker of immune activation at baseline is also a negative predictor of successful virologic response to pegIFN/RBV therapy during HCV-HIV coinfection. It is conceivable that immune dysfunction associated with immune activation is responsible for this correlation. It is also conceivable that factors underlying the observed correlation between baseline HCV level and sCD14 may contribute. However, although baseline HCV load is regarded as a predictor of virologic response [9], this was not the case in the present study, suggesting sCD14 may independently predict viral decline in patients coinfected with HIV. Similarly, sCD14 has been associated with fibrosis, aspartate aminotransferase to platelet ratio index (APRI), and HCV therapy outcome during HCV monoinfection [5]. However, we did not observe an association between APRI and sCD14, or APRI and viral decline magnitude here, again suggesting that sCD14 may independently predict viral decline and perhaps even serve as a useful biomarker. We also found baseline sCD14 to associate with baseline CD16⁺CD56⁻ NK subset frequency, and baseline CD16⁺CD56⁻ NK-subset frequency to negatively associate with HCV decline. This latter finding confirms and extends our prior finding of this expanded NK-cell subset negatively correlating with HCV therapy outcome [10], now linking frequencies of these cells in the peripheral blood to levels of sCD14 in HIV-HCV coinfection. Conceivably, there may be linked mechanisms of immune activation governing levels of sCD14 and frequencies of these NK cells. Again, interferon-mediated activation may be one common factor, and previous data evaluating NK subset CD69 expression as a marker of activation support of the concept that lower baseline NK-cell activation may predict greater responsiveness to IFN-α and better therapy outcome [15]. Further analysis in larger cohorts is required to clarify this relationship.

Genetic variations in the IL28B gene (rs12979860 and rs8099917) have been associated with pegIFN/RBV SVR in patients with genotype-1 HCV infection [7]. These SNPs are not in linkage disequilibrium, and we observed an association with viral clearance for rs12979860 but not rs8099917. The lack of association with rs8099917 may be due to small sample size, a weaker relation between this SNP than rs1297860 and viral clearance, or perhaps differences in population structure between studies.

In summary, our data indicate that a pretreatment marker of monocyte activation is positively correlated with HCV level and negatively associated with viral kinetics and outcome of pegIFN/RBV therapy during HCV-HIV infection. Furthermore, we find a similar negative relationship between CD16⁺CD56⁻ NK-cell levels and treatment response and also indicate an association between the frequency of this NK-cell subset and monocyte activation. Together, these findings support the notion that the status of the innate cellular immune system influences the ability to respond to HCV treatment.

After manuscript acceptance, the authors note additional newly published evidence of sCD14 level associating with pegIFN/RBV-therapy outcome [16].

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/)). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.