The role of socio-behavioral factors in sub-protective tenofovir diphosphate (TFV-DP) levels among YMSM enrolled in two PrEP trials

Abstract

Background

Young men who have sex with men (YMSM) experience disparities in HIV acquisition more than any other group. Daily oral pre-exposure prophylaxis (PrEP) with tenofovir/embricitabine (TDF/FTC) has been shown to effectively prevent HIV transmission in YMSM; however, recent studies suggest that young Black men who have sex with men (YBMSM) experience sub-protective levels of tenofovir diphosphate (TFV-DP) more frequently than other groups.

Setting

Combined data from Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) 110/113, two open label PrEP studies that provided PrEP and evidence-based behavioral interventions to young MSM (YMSM) age 15 to 22 years old.

Methods

Bivariate and logistic regression analysis were used to examine sociodemographic and behavioral factors associated with protective TVF-DP levels (defined as ≥700 fmol/punch) in ATN 110/113 data.

Results

In bivariate analysis, self-identified Black participants, residential displacement due to sexual orientation, low perceived risk, and stigma with the medication were associated with sub-protective levels. Hispanic ethnicity was associated with protective levels. In the final models, Black males were less likely to have sub-protective levels than non-Black males at 4, 8, 12 weeks. Self-reported displacement due to sexual orientation was associated with sub-protective levels while older age was as associated with protective levels.

Conclusion

These findings highlight how future behavioral research and biomedical prevention efforts in YMSM will need to address PrEP disparities that may occur in YBMSM, perception of risk and lack of key supportive housing during this period that may be critical factors that contribute to HIV acquisition.

Background

Young men who have sex with men (YMSM) disproportionately are impacted by HIV, with young Black men experiencing the greatest burden of HIV disparities.^1,2 Daily oral pre-exposure prophylaxis (PrEP) with tenofovir/embricitabine (TDF/FTC) has been shown to effectively prevent HIV transmission in YMSM;^3,4 however, recent studies suggest that young Black men who have sex with men (YBMSM) experience sub-protective levels of tenofovir diphosphate (TFV-DP) more frequently than other groups, potentially predisposing them to HIV acquisition despite use.^5,6

TFV-DP levels greater than 4 pills per week [≥700 fmol/punch] have been associated with high levels of protection against rectal HIV exposure.⁷ Data from the Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) 110/113 studies demonstrated a decline in adherence over the course of the 48-week follow-up in all participants, with sub-protective levels beginning to occur around 12 weeks. YBMSM, however, had sub-protective levels at all time points, with median levels below the protective threshold of 4 pills per week [<700 fmol/punch].⁵ There is a lack of clarity as to why YBMSM have sub-protective TFV-DP levels compared to other groups; however, prior work suggests that the lower levels are not attributed to biological differences.⁸

Socio-behavioral factors, that predispose all YMSM to HIV risk, such as low perceived HIV risk and stigma, may be further compounded by unstable housing and poor family support, further contributing to medication non-adherence and lower protective TFV-DP levels in YBMSM.⁹ The objectives of this study were to examine sociodemographic and behavioral factors associated with protective TVF-DP levels (defined as >700 fmol/punch).

Methods

Combined data from ATN 110/113, two open label PrEP studies that provided PrEP and evidence-based behavioral interventions to young MSM (YMSM) age 15 to 22 years old, were analyzed to examine the objective of this study. In these studies, participants were followed through study visits at baseline and monthly for 3 months, then quarterly through week 48. At each visit, participants completed behavioral assessments that included reasons for and frequency of missing PrEP via an audio computer-assisted self-interview (ACASI), received condoms, and were provided study drug.¹⁰ Participants were provided compensation for each study visit as determined by the IRB at each site (range $50–75). A full description of the studies is published elsewhere.^5,6

Dried blood spots collected at each visit were used to quantify intracellular TFV-DP concentrations in red blood cells.¹¹ For the purpose of this analysis, primary outcome was defined as protective (≥700 or more fmol per punch) and sub-protective levels defined as a TFV-DP level <700 fmol/punch.

The main predictor examined of protective level was Black/African American versus non-Black race. We additionally examined potential factors associated with optimal TFV-DP levels including age (younger: 15–18 years vs. older: 19–22 years); Hispanic ethnicity; housing status (living in or outside the presence of parents/guardian, history of being kicked out of the home due to sexual orientation); condom non-use, substance use, perceived risk of acquiring HIV and self-reported reasons for discontinued use. Age group categories were based on sample size, so that each group would have approximately the same number of participants. Self-reported reasons for discontinued use were dichotomized (no/yes) from a four-point scale (1=Never, 2=Rarely, 3=Sometimes, and 4=Often) as the majority of the subjects indicated never discontinuing use (83.1% to 97.0%).

Bivariate analyses were used to explore associations between the main predictor variable and potential variables that might impact adherence at 4, 8, 12 and 24-week follow-up time points in separate models. Variables significant (p<0.10) in the bivariate analyses for each time point were included in logistic regression models. Interaction terms were explored between age and race. The four forward stepwise regression models (week 4, 8, 12, and 24) included race, age, and the specific significant bivariate covariates for that specific time point.^† We further explored factors associated with optimal adherence at each time point in YBMSM only and by age to examine specific factors that might impact adherence in YBMSM and whether factors significant at each time point mediated the relationship between race and blood levels.

To adjust for multiple comparisons across time, we used the Bonferroni correction for the logistic regression models to determine significant predictors [p-value < 0.0125 (α=0.05/4 time points)].

Results

Participant characteristics are presented in Table 1. At baseline, the mean age of the sample was 19.12 ± 2.01 years. Most participants self-identified as African-American/Black (55%), were living with parents/guardians (47%) and described prior alcohol (82%) or marijuana use (55%). One-fifth (19.2%) of participants described STI in the past year, being displaced due to sexual orientation (18%) and over one-third described condom non-use during receptive (38.9%) or insertive (38.1%) sex. Overall, sub-protective TFV-DP levels (51.1%) were common, with Black participants having TFV-DP levels at all time points that were sub-protective (range 495.90–649.70).

Table 1:

Overall Variables by TFV-DP levels

	Overall	4 weeks			8 weeks			12 weeks			24 weeks
		TFV-DP <700 [%(N)], M (SD)  59.3% (140)	TFV-DP ≥700 [%(N)], M (SD)  40.7% (96)	P value	TFV-DP <700 [%(N)], M (SD) 43.6% (99)	TFV-DP ≥700 [%(N)], M (SD) 56.4% (128)	P value	TFV-DP <700 [%(N)], M (SD) 44.3% (97)	TFV-DP ≥700 [%(N)], M (SD)  55.7% (122)	P value	TFV-DP <700 [%(N)], M (SD) 57.1% (116)	TFV-DP ≥700 [%(N)], M (SD)  42.9% (87)	P value
Age (MD ±SD)	19.12 ± 2.01	19.21±2.00	91.11±2.03	0.729	19.01±1.98	19.31±2.04	0.263	18.98±1.95	19.37±2.06	0.156	18.84±2.01	19.71±1.92	0.002
15–18	37.2% (103)	59.8% (52)	40.2% (35)	0.915	48.8% (40)	51.2% (42)	0.238	51.3% (40)	48.7% (38)	0.121	70.4% (50)	29.6% (21)	0.005
19–22	62.8% (174)	59.1% (88)	40.9% (61)		40.7% (59)	59.3% (86)		40.4% (57)	59.6% (84)		50.0% (66)	50.0% (66)
Race					-	-						-	-
Black	55.0% (144)	71.3% (92)	28.7% (37)	<0.001	50.8% (61)	49.2% (59)	0.017	55.6% (65)	44.4% (52)	0.001	64.3% (72)	35.7% (40)	0.048
Non-Black	45.0% (118)	44.1% (41)	55.9% (52)		34.4% (32)	65.6% (61)		31.5% (28)	68.5% (61)		50.0% (40)	50.0% (40)
Hispanic	31.4% (87)	52.1% (38)	47.9% (35)	0.098	40.5% (30)	59.5% (44)	0.470	33.3% (23)	66.7% (46)	0.028	54.7% (35)	45.3% (29)	0.610
Kicked out due  to Sexual  Orientation	18.6% (51)	68.2 (30)	31.8 (14)	0.18	59 (23)	41 (16)	0.035	73(27)	27(10)	<0.001	74.3(26)	25.7(9)	0.023
Discontinue  medication  due to  depression	-	77.8 (21)	22.2 (6)	0.037	64(16)	36(9)	0.02	56.5(13)	43.5(10)	0.188	68.4(13)	31.6(6)	0.28
Discontinue  medication  due to low  perceived  risk	-	93.8 (15)	6.3 (1)	0.004	71.4(10)	28.6(4)	0.02	85.7(12)	14.3(2)	0.001	92.9(13)	7.1(1)	0.005
Discontinue  medication  due to pills  viewed as  toxic	-	90.9 (10)	9.1 (1)	0.029	80(8)	20(2)	0.019	75(6)	25(2)	0.08	75(6)	25(2)	0.47
Discontinue  medication  due to fear  that others  might see	-	81.3 (13)	18.8 (3)	0.063	73.3(11)	26.7(4)	0.012	76.9(10)	23.1(3)	0.013	91.7(11)	8.3(1)	0.012

In bivariate analysis (Table 1), self-identified Black participants were less likely to have protective TFV-DP levels at all time points. At 4 weeks, persons who reported not taking medications because of self-reported feeling depressed/overwhelmed, not engaging in risky sex, and belief that medications were toxic/harmful, were less likely to have protective TFV-DP levels than participants who did not report these reasons. At 8 weeks, history of being kicked out of their home, not taking medications because of feeling depressed/overwhelmed, not engaging in risky sex, belief that medications were toxic and desire for others not to see medications, were less likely to have protective TFV-DP levels than participants who had not been kicked out or did not report these reasons. Hispanic participants were more likely to have protective levels at 12 weeks than non-Hispanic participants. At 12 and 24 weeks, history of being kicked out of their home, not engaging in risky sex, and desire for others not to see medications, were less likely to have protective TFV-DP levels than participants who had not been kicked out or did not report these reasons. Age was significant at 24 weeks, with older participants more likely to have protective levels than younger participants. An interaction between age and race was seen at 8 weeks with older (19–22 years) non-Black participants more likely to have protective blood levels than younger Black participants and older (age 19–22) Hispanic participants at 4 and 8 weeks more likely to have protective levels than non-Hispanic participants of the same age. However, no difference was noted in level of protection for Hispanic Black participants.

Using logistic regression modeling, at 4 and 8 weeks, Black participants were less likely to have optimal protective levels than non-Black participants (68 and 55 percent respectively, Table 2). At week 12, Black participants were 59 percent less likely than non-Black participants to have protective levels. Participants who reported being kicked out of the house were 81 percent less likely than participants who were not kicked out of the house to have protective levels. Older participants were 1.97 times more likely to have protective levels than younger participants. At week 24, older participants (ages 19–22) were 2.4 times more likely than younger participants (ages 15–18) to have protective levels. Perception of risk also approached significance at 12 and 24 weeks with participants who reported missing their medication because they were not having risky sex 91.6 and 92 percent (OR, 95% CI, p-value 0.022; OR, 95% CI, p-value 0.016, respectively) being less likely than participants who did not report this reason to have protective TFV-DP levels.

Table 2:

Variables associated with ≥ 700 TFV-DP levels

	4 weeks			8 weeks			12 weeks			24 weeks
Overall	Odds Ratio	p-value	95%CI	Odds Ratio	p-value	95%CI	Odds Ratio	p-value	95%CI	Odds Ratio	p-value	95%CI
Older Age (19–22)										2.705	0.002	0.010–0.628
Black Race	0.324	<0.001	0.182–0.577	0.455	0.008	0.254–0.815	0.381	0.003	0.202–0.718	-	-	_
Kicked Out	-	-	-	-	-	-	0.212	0.001	0.086–0.522		-	_
Depressed	-	-	-	0.413	0.061	0.164–1.041	-	-	-	-	-	_
Black Participants Only
Kicked Out	-	-	-	-	-		0.131	0.002	0.037–0.470	-	-

Note: Adjusted p-value for statistical significance is 0.0125

In the logistic regression models examining factors associated with protective levels in Black participants only (Table 2), at 12 weeks, Blacks who described having been kicked out were 86.6% less likely to have optimal TFV-DP adherence compared to those who reported not being kicked out. No variables were significant at any of the time points in the logistic regression models of younger (ages 15–18) Black participants. In older (ages 19–22) Black participants, having been kicked out of one’s home was significant at 12 weeks, indicating that those who were kicked out were 95.2% less likely to have protective levels than those who reported they were not kicked out (p=0.005). At 24 weeks, self-reported displacement nears significance in the same direction with a p-value of 0.027. No factors significant at each time point mediated the relationship between race and blood levels.

Conclusions

In this evaluation of ATN 110 and 113 data, we found that self-identified YBMSM were less likely to have therapeutic levels of TFV-DP across all data points. Potential vulnerabilities, including displacement due to sexual orientation, low perceived risk, and stigma associated with medications, impacted optimal adherence and need to be critically addressed in order to implement PrEP in adolescents.

Nearly twenty-percent of youth reported having been displaced due to sexual orientation. Residential instability has been associated with HIV risk-related behaviors including, higher number of sexual partners and history of STIs.¹² Other studies have described an association between residential displacement and vulnerability, particularly among young men of color, whereas family connectedness and support have been protective.^13–15 Displacement may be even more critical for YBMSM who may be already experiencing isolation, marginalization and homophobia within Black communities and across other sexual and gender minority communities.¹⁶ One approach is to develop interventions specifically for parents and families of young gay, lesbian, bisexual, or questioning youth in order to ameliorate the conflict that can arise as an adolescent begins to explore their same-sex sexual attractions.¹² This work also calls for policy measures that expand housing for young men who may experience residential displacement and instability as a result of their sexual orientation due to the high risk for HIV acquisition during displacement.

Older participants were 2 times more likely to have therapeutic levels than younger participants. This finding suggests that younger adolescents have some specific developmental and cognitive needs that are required to successfully navigate PrEP use. Some have suggested that younger adolescents will require not only additional visits than what is currently recommended by national guidelines, but also team members that can address these needs and help youth incorporate PrEP into their daily lives.¹⁷ The ability to develop abstract thinking, plan for one’s future, and integrate a positive adult identity, while simultaneously managing potential disapproval of same-sex behavior, and daily medications for HIV prevention, is likely to require an interdisciplinary developmentally-centered approach needed to support optimal adherence.

In the United States, rates of HIV acquisition in adolescents are highest in YMSM.^18,19 YBMSM experience some of the greatest health disparities, with studies suggesting that 40% of YBMSM will acquire HIV by 40 years old unless prevention efforts improve.²⁰ The health disparities inherent in the U.S. epidemic are heightened in the what we know about TDF/FTC uptake in Black men who have sex with men.²¹ In 2012–2015, only 11.5% of the prescriptions filled by youth under 25, were prescribed to non-Hispanic Black users. ²² Self-reported use in YBMSM samples has been estimated at closer to 8% in YBMSM ²³ and recent work suggests that YBMSM experience the lowest rates of uptake,²⁴ despite having greatest indications for PrEP. ²¹

Stigma associated with the medication and low risk perception have been identified as key barriers to PrEP use.^5,6,25 We found that adherence was associated with not engaging in “risky sex” and fear others would see the medications. Confidentiality concerns about PrEP, including disclosure of one’s sexual behavior and sexual orientation to parents/guardians has been described as a barrier to YMSM ²⁶ and may further prevent access to PrEP for youth who are most vulnerable.

In order for biomedical interventions to be effective, researchers, health care providers, and policy makers will need to better understand and address how PrEP is received in at-risk YMSM; and to develop approaches that address such barriers. Prior work suggests that PrEP is perceived as promoting “risky sexual practices” or only for individuals engaging in sex with multiple partners.²⁵ This assumption excludes persons in serodiscordant relationships and focuses on individual risk behavior, leaving out partners who may be located in networks of high HIV prevalence.²⁷ The findings of this study calls for work that explores ways in which YMSM process sexual risk and develops interventions that promote uptake of PrEP, addresses the stigma of taking a medication for HIV, and provide needed protections (e.g., access to needed housing) during disclosure of sexual orientation.

There are some limitations of this data that should be noted. Both ATN 110/113 datasets were small open label studies and may not reflect how YMSM apply and use PrEP in their daily lives. The size of the datasets may have limited our ability to identify associations. Despite these limitations, the findings of this work suggest key areas of intervention for PrEP in YMSM.

Implications.

These findings have important implications with respect to both future behavioral research and biomedical prevention efforts. Research is needed to better understand why suboptimal adherence may be a relatively common occurrence among YMSM who are attempting to adhere to daily PrEP use, and why some populations, particularly Black and younger MSM may be especially vulnerable to sub-protective levels. Effective PrEP interventions will need to address not only the developmental, cognitive and risk perception of the individual, but also incorporate the family support and community contexts that may further promote risk for YMSM. Particularly, the lack of key supportive housing during this period may be one critical factor that will need to be addressed in future PrEP interventions.