Prescribing tenofovir/emtricitabine (TDF/FTC) as PrEP raises questions about side effect risks in HIV-negative people. Side effects that arise in the first weeks of PrEP -- nausea, abdominal cramping, vomiting, dizziness, headache, and fatigue -- usually resolve without withdrawing TDF/FTC. But providers should alert PrEP candidates to these possible problems so they do not stop PrEP unnecessarily if side effects occur. Cohort studies that enroll HIV-positive people and similar HIV-negative people at risk of HIV infection offer unique insights into risks of two well-known TDF side effects, impaired kidney function and declining bone mineral density (BMD). This research indicates that many HIV-negative PrEP candidates have risk factors for kidney and bone complications, but these people do not appear to have higher than normal rates of kidney and bone problems in published studies. Through 1 to 2 years of follow-up, placebo-controlled PrEP trials found higher rates of declining kidney function and BMD in HIVnegative people randomized to TDF/FTC. But these problems affected only small proportions of people assigned to TDF/FTC during these trials. In HIV-positive people, some research indicates that TDF-linked kidney toxicity improves when TDF stops, but a 10,000-person Veterans Affairs study found that it may not. Another prospective study of HIV-positive veterans figured that every year of TDF use boosts the risk of osteoporotic fracture 12%, while other cohort studies and trials confirmed dwindling BMD with TDF therapy but found no greater fracture risk. TDF/FTC PrEP should not be prescribed for people with creatinine clearance below 60 mL/min, and pre-PrEP DEXA scans of BMD may be prudent for people with a past fracture or other bone risk factors.
"I probably wouldn't take [PrEP] because I know HIV medications are very strong and if you don't have to take them why would you? And I'm healthy, so why would I do damage to my body to protect myself but I still got a chance of getting [HIV], when I can just use a condom and continue what I've been doing?"1
Those keen questions by a gay man in Los Angeles, reported in a study by UCLA investigators,1 neatly encapsulate the dilemma faced by people who may take tenofovir/emtricitabine (TDF/FTC) preexposure prophylaxis (PrEP) and by people who may prescribe it. Plenty of precedent sketches out the risks healthy people face when taking prophylactic agents -- from short-term chloroquine to ward off malaria to regular hormonal agents to prevent pregnancy.
Hormonal contraception, in fact, offers a trenchant reverse analogy for TDF/FTC PrEP. Hormonal contraception, especially injected hormonal agents, may boost the risk of HIV acquisition and transmission.2 But because data on that question are inconsistent and contraception offers such clear family-planning benefits, the World Health Organization decided that women using progestogen-only injectable contraception "should be strongly advised to also always use condoms, male or female, and other HIV preventive measures."3
In contrast, TDF/FTC PrEP clearly cuts the risk of HIV acquisition,4-6 but it does so at the risk of TDFrelated toxicity and so requires pre-PrEP side-effect screening and regular check-ups during PrEP (see Table 2 in the first article in this issue). Like hormonal contraceptives, TDF/FTC PrEP should also be taken "in combination with safer sex practices,"7 because it does not erect an impervious firewall against HIV.
Probably the most hallowed tenet of medicine -- and one echoed often at the FDA hearing on TDF/FTC PrEP -- adjures clinicians trying to help their patients to avoid hurting them first. A PrEP-specific translation of that dictum, penned by Myron Cohen and Lindsey Baden (University of North Carolina and Brigham and Women's Hospital), puts it this way: "Providing a daily medication to healthy, HIV-uninfected persons demands an extraordinarily high degree of safety."8
Emtricitabine ranks among the safest medicines dispensed, but everyone knows TDF poses a threat of serious long-term toxicity to a handful of people taking it to treat HIV infection. Will people without HIV face the same risk?
Because follow-up in PrEP trials generally falls short of 2 years, only informed guesswork can approach an answer to that question. But thanks to round-the-clock HIV research, the guesswork addressing this question is highly informed. Over the years, HIV investigators created and endlessly canvassed cohorts of HIV-positive people paired with HIV-negative contemporaries sharing similar behaviors, lifestyles, sociodemographic nitty-gritties, and -- in the end -- a worrying risk of HIV infection. These well-studied HIV-negative people are among today's top PrEP candidates. The Multicenter HIV Cohort Study (MACS) of US gay and bisexual men with and without HIV and the Women's Interagency HIV Study (WIHS) of US women are eminent examples. PrEP studies of HIV-negative people who run a high risk of getting infected provide another font of plumbable data.
Together, data from this research offer much more than a passing glance at rates and risks of two TDFrelated toxicities in HIV-negative people: impaired kidney function and waning bone mineral density (Figure 1). This article offers a detailed analysis of kidney and bone findings in people without HIV -- as well as in people with HIV who took TDF for years.
Short-Term Side Effects of TDF/FTC PrEP
Although providers worry about the long-term risk of compromising kidneys or depleting bone mineral with TDF/FTC PrEP taken for 2 years or more, both clinicians and PrEP candidates should realize that TDF/FTC can have short-term side effects, including nausea, abdominal cramping, vomiting, dizziness, headache, and fatigue.4-6,99 These problems typically arise in the first week or 2 of PrEP, then often disappear in the next few weeks. In an interview in this issue of RITA! Robert Grant, who headed the iPrEx PrEP trial,4 compares these problems to the "start-up syndrome" seen when people start antiretrovirals for treatment. These vexations can have a big impact on care if they sway people to skip doses or stop their drugs altogether.
Clinicians would be wise to alert PrEP novitiates that these ailments may arise and that they usually resolve without stopping TDF/FTC. Robert Grant even suggests that providers schedule a "check-in visit" 2 to 4 weeks after a person plans to start PrEP -- both to see if the person actually started taking the pill daily, and if TDF/FTC caused any aches or pangs.
FDA prescribing information for TDF/FTC lists three adverse reactions reported in at least 2% of people randomized to TDF/FTC PrEP and more frequently than in those randomized to placebo -- headache, abdominal pain, and weight loss.7 A few other problems affected at least 2% of TDF/FTC takers and a similar proportion of placebo recipients: diarrhea, back pain, depression, and anxiety. "Unintentional weight loss" of 5% or more affected 34 people in the TDF/FTC arm of iPrEx and 19 in the placebo group (P = 0.04).4
An Angle on Long-Term Side Effect Risk With TDF/FTC PrEP
How healthy are HIV-negative people who may take TDF/FTC PrEP in the United States? And what's their risk of subpar kidney function and ebbing bone mineral density? Because many sexually active people -- and all injection drug users -- run some risk of picking up HIV, it's impossible to generalize about at-risk people as a monolithic group. But the Multicenter AIDS Cohort Study (MACS), the Women's Interagency HIV Study (WIHS), and other populationbased analyses in high-income countries offer plenty of insight on HIV-negative gay and bisexual men and women in danger of getting HIV infection.
By and large, people who put themselves on a collision course with HIV are not paragons of good health. Because people with the highest risk of picking up HIV during sex are those who have lots of sex -- often without condoms -- this group shoulders a high burden of other sexually transmitted infections, including hepatitis C virus (HCV) infection, an oftnoted risk factor for deliquescing bone density and chronic kidney disease. In the United States, young black men who have sex with men (MSM) account for a burgeoning proportion of new HIV infections,10 and blacks run a higher risk of kidney disease than whites. Young white and black MSM with lots of sex partners often have other habits that threaten their health -- smoking, drinking, and downing recreational drugs that range from the innocuous to the caustic. In the United States, WIHS findings and other data indicate, women with a high HIV risk are often poor, overweight, and members of minorities with off-andon access to health care.
If one compiles a list of classic risk factors for low bone density and chronic kidney disease (Table 1), that catalog includes an array of variables common among US men and women with a substantial risk of HIV infection. Both MSM and high-risk women can claim many of the behavioral risk factors. HCV and other chronic infections occur often among risktaking gay and bisexual men, while disadvantaged women with a high HIV risk are prone to diabetes, hypertension, high cholesterol, and cardiovascular disease. Unbalanced diets in at-risk poor women can result in vitamin D and calcium deficiency. Vitamin D insufficiency and deficiency are virtually endemic in sedentary populations with meager sun exposure.
Table 1. Classic Risk Factors for Low Bone Mineral Density and Chronic Kidney Disease
The TDF/FTC PrEP dose is the same as the treatment dose -- one Truvada tablet daily, 300 mg of TDF plus 200 mg of FTC. Whether PrEP users will adhere to that prescription remains open to question (as discussed in the preceding review article in this issue). But if they take TDF/FTC PrEP for 2 or 3 years, they may run the same toxicity risk as people who take TDF/FTC daily for chronic HIV infection -- though perhaps not quite. People with HIV may face a slightly higher risk of side effects because HIV itself may affect kidney function and bone density, because HIV-positive people may have higher rates of other conditions that threaten kidneys and bone, and because HIV-positive people may be taking more nonantiretrovirals that pose toxic threats to kidneys and bones. So reviewing the impact of TDF/FTC in HIV-negative people enrolled in PrEP trials -- and appraising kidney and bone health in other PrEP candidates -- should yield some insight into long-term toxic risks with this double drug. After those analyses, this article weighs evidence on long-term TDF toxicity in people taking antiretroviral therapy.
Kidney Function in PrEP Users and Candidates
FDA regulators recognized the kidney threat posed by long-term TDF when they stipulated that TDF/FTC PrEP should not be used by "HIV-uninfected individuals if creatinine clearance is below 60 mL/min."7 In iPrEx,4 Partners PrEP,5 and TDF2,6 PrEP users took TDF -- often inconsistently -- for medians of 1.2, 1.9, and 1.1 years in the primary published reports, and TDF did provoke some renal wrinkles in these HIV-negative study participants (Table 2). In the iPrEx trial of MSM and transgender women, investigators recorded 41 creatinine elevations during follow-up, 26 (2%) in people randomized to TDF/FTC and 15 (1%) in those randomized to placebo, a difference approaching statistical significance (P = 0.08).4 Eighteen creatinine elevations (44%) remained in the normal range, and 36 elevations (88%) disappeared on the next test. Five people randomized to TDF/FTC (5%) and none randomized to placebo had creatinine jumps on more than 1 consecutive test. Ten creatinine gains led iPrEx participants to stop study drug, 7 in the TDF/FTC group and 3 in the placebo group. Nine people resumed their assigned regimen.
Table 2. Key Findings on Kidney Health in HIV-Negative at-Risk People Taking or Not Taking TDF
Transaminase elevations affected 2% of iPrEx participants randomized to TDF/FTC and 1% randomized to placebo.4
Ten creatinine elevations -- 7 in the TDF/FTC group -- led iPrEx participants to stop their study drug. Nine resumed their assigned regimen.4
Four Partners PrEP participants randomized to TDF or TDF/FTC and 1 randomized to placebo stopped study drug because of declining creatinine clearance, but overall creatinine and phosphorous abnormalities did not differ between study arms.5
Renal side effect rates in PrEP trials must be interpreted in the context of the relatively low adherence reported in these trials.
Creatinine clearance was normal in gay and bisexual HIV-negative at-risk men in a MACS study, but 37% had hypertension and 13% had diabetes, risk factors for chronic kidney disease.12
Among HIV-negative at-risk women in the HERS cohort, 6.6% had renal lab abnormalities, compared with 21.7% of HIV-positive women in that study.15
Among HIV-negative at-risk women in HERS, many had two chronic kidney disease risk factors: 47.5% tested positive for HCV and 24.2% had hypertension.15
HERS, HIV Epidemiology Research Study; MACS, Multicenter AIDS Cohort Study.
In the Partners PrEP trial of HIV-discordant African couples, frequency of creatinine or phosphorus abnormalities did not differ between the TDF arms and the placebo arm.5 Three people taking TDF alone and 2 taking TDF/FTC had grade 2 or 3 creatinine elevations, with both frequencies below 1%. Four Partners PrEP participants stopped TDF or TDF/FTC because creatinine clearance dropped below 50 mL/min, while 1 stopped placebo because of declining clearance. Seven Partners PrEP participants permanently stopped their assigned regimen, 6 of them because of grade 2 renal toxicity (3 taking TDF alone, 2 taking TDF/FTC, and 1 taking placebo). Grade 2 and 3 drops in phosphorus affected 8% and 1% taking TDF alone and 8% and 1% taking TDF/FTC.
In the TDF2 study of high-risk heterosexual African women and men, 1 person (0.2%) randomized to TDF/FTC and none randomized to placebo had a creatinine elevation, while 23.2% randomized to TDF/FTC and 26.2% randomized to placebo had low phosphorus.6
A phase 2 trial of TDF alone versus placebo to prevent HIV infection in high-risk women of Cameroon, Ghana, and Nigeria collected primary safety data from two sites.11 Comparing 363 women randomized to TDF and 368 randomized to placebo for 210.2 and 217.6 person-years of follow-up, researchers recorded no grade 2, 3, or 4 creatinine elevations in either study arm. There were 13 grade 1 elevations in the TDF group (6.5 per 100 person-years) and 15 in the placebo group (7.1 per 100 person-years). Phosphorus drops also proved uncommon and did not differ between groups.
Together these findings suggest that TDF taken as PrEP for 1 to 2 years poses only a small risk of negative kidney marker changes, and by some (but hardly all) measures that risk was greater with TDF or TDF/FTC than with placebo.
What about kidney health in people not taking TDF but with more than a passing risk of HIV infection, in other words, PrEP candidates? Multiple PubMed searching strategies turned up two reports of kidney function in such people, one in men and one in women (Table 3).
Table 3. Kidney Risk Factors in HIV-Negative but at-Risk Men and Women
Men in MACS12 (n = 150)
Women in HERS15 (n = 425)
One third black
Hypertension in 37%
Antihypertensive therapy in 24.2%
Diabetes in 13%
HCV in 47.5% Renal lab abnormalities in 6.6%
HERS, HIV Epidemiology Research Study; MACS, Multicenter AIDS Cohort Study.
The study in men involved 738 HIV-positive and 150 HIV-negative gay and bisexual men in MACS.12 MACS is an ongoing prospective study of gay and bisexual men with and without HIV infection recruited in Baltimore, Chicago, Pittsburgh, and Los Angeles.
The 6972 cohort members, including 3501 men with HIV, make study visits twice a year. This cross-sectional analysis involved men with serum creatinine and urine protein excretion measured between September 2006 and April 2007. One third in each group was black, and 2% or fewer injected drugs. The HIVpositive and negative men had similar chronic kidney disease risk factors, except that a higher proportion of HIV-positive men had HCV infection (10% versus 4%) and high liver enzymes. Proportions with diabetes were similar in the HIV-positive and negative groups (10% and 13%), while more than one third in each group (38% and 37%) had hypertension, another kidney disease risk factor.
A significantly higher proportion of men with than without HIV had proteinuria (17% versus 2%, P < 0.01). Median creatinine was similar in the two groups (1.0 mg/dL without HIV and 0.9 mg/dL with HIV, P = 0.41), while median cystatin C, an alternative kidney function marker, was significantly higher in the HIV group (0.76 versus 0.85 mg/dL, P < 0.01). Median estimated glomerular filtration rate (eGFR) using serum creatinine was similar in the two groups (103.1 mL/min without HIV versus 105.2 mL/min with HIV, P = 0.78), while eGFR using cystatin C was significantly greater in men without HIV (108.0 versus 94.6 mL/min/173m2, P < 0.01).
The National Kidney Foundation lists normal eGFR as above 60 mL/min,13 while labtestsonline.org gives 90 to 120 mL/min as normal.14 Either way, both HIVpositive and HIV-negative men in this MACS analysis fell within the normal range as a group. Interquartile ranges also fell entirely within the 90-to-120 normal range for both groups of men with both eGFR estimating formulas.
The HIV Epidemiology Research Study (HERS) prospectively monitored HIV-positive women and HIVnegative women with a high HIV risk recruited in Baltimore, the Bronx, Detroit, and Providence.15 An analysis involving 885 HIV-positive women and 425 without HIV who made twice-yearly visits from 1993 through 2000 tracked overall and condition-specific hospital admissions, including admissions for diabetes mellitus and nonacute renal conditions. About two thirds of women in the positive and negative groups were between 31 and 44 years old; 61% of positive women and 53% of negative women were black, and 17% and 15% were Hispanic.
While 21.7% of HERS women with HIV had renal lab abnormalities, 6.6% of HIV-negative women did, a highly significant difference (P < 0.0001).15 The hospital admission rate for nonacute renal causes was more than 14 times higher in women with than without HIV (rate ratio 14.4, P = 0.0033). Diabetes admissions were 3.6 times more likely in women with than without HIV (P = 0.04). Overall admission rates were 54.9 per 100 person-years in the HIV group and 15.1 per 100 person-years in the HIV-negative group. In 2009 the hospital discharge rate for all US women was 13.8 per 100 person-years, according to the CDC.16 So the overall admission rate for HIVnegative women in HERS is largely in line with the general-population rate.
Although these HERS findings indicate that US women at risk of HIV infection have lower rates of kidney problems and kidney-related and overall hospital admissions than women with HIV, other findings show that both groups had high rates of two renal risk factors. About one quarter of women with and without HIV (27.0% and 24.2%) were taking antihypertensives, a nonsignificant difference. Hypertension quintupled the risk of nonacute renal hospital admissions in women with HIV (rate ratio 5.1, 95% CI 2.1 to 12.4); the researchers did not perform a similar analysis for HIV-negative women. While 61.4% of HIV-positive women tested positive for HCV antibody, 47.5% without HIV had a positive HCV antibody result (P < 0.0001). Drug injection probably accounted for the high HCV prevalence, as 60.7% of HIV-positive women and 54.1% of HIV-negative women had an injection history.
Together these two studies suggest that US men and women at risk of HIV infection do not have more compromised kidney function than the general population. However, the high hypertension prevalence in MACS men and HERS women without HIV offers a vivid example of kidney threats faced by prime PrEP candidates. The same can be said for the high HCV-antibody positivity rate in HIV-negative HERS women.
Kidney Changes With Long-Term TDF Therapy
The medical literature is loaded with studies of how TDF does -- or does not -- affect kidney function when taken for several years by people with HIV (Table 4). Because of differences in study populations, design, and follow-up time, these analyses yield varying, and sometimes seemingly contradictory, results.
Table 4. Key Findings on Kidney Health in HIV-Positive People on Long-Term TDF
In a prospective 10,841-person US veterans cohort, every year of TDF therapy independently raised the risk of proteinuria, rapid kidney function decline, and chronic kidney disease, and these changes were not readily reversible after TDF stopped.17
A Spanish study of 183 people with TDF-associated kidney impairment found that kidney markers improved in 69% after TDF stopped, including 59% in whom key markers returned to normal.18 An 80-person US study charted improving kidney function in 76% of people who stopped TDF.19
A US multicenter study of 3329 people determined that TDF plus a ritonavir-boosted protease inhibitor (PI) (but not TDF alone or a boosted PI alone) independently raised the odds of eGFR below 60 mL/min, which it did in 5.7% of patients.20
A EuroSIDA analysis of 6843 people determined that TDF and three PIs each independently raised the risk of chronic kidney disease.25
Meta-analysis of studies comparing TDF regimens with non-TDF regimens found an average 3.92 mL/min lower creatinine clearance in TDF takers, along with a 0.7% higher risk of acute kidney failure.22
A UK study of 3439 people with HIV found that, among those who started TDF, being 50 or older boosted the odds of chronic kidney disease 5.4 times, while an eGFR of 60 to 75 raised the odds 17.2 times.26
The biggest TDF-kidney study involved 10,841 HIVpositive US veterans, almost all of them (98%) men.17 While 40% of this nationally representative sample had taken TDF, 60% had not. Median age in this prospective study was similar in TDF takers and TDF naives (45 and 47), and similar proportions were black (47% and 51%). Among kidney risk factors, baseline hypertension prevalence was 39% without TDF experience and 38% with TDF experience, diabetes prevalence 7.9% and 6.8%, abnormal lipids 15% in each group, HCV positivity 17% and 14%, proteinuria 21% and 19%, and smoking prevalence 19% and 18%.
The VA team examined associations between TDF use and time to three kidney endpoints: (1) proteinuria (two consecutive urine dipstick measurements at or above 30 mg/dL), (2) rapid decline in kidney function (at least 3 mL/min annual decline), and (3) chronic kidney disease (eGFR below 60 mL/min). Median follow-up before an endpoint was reached stretched from 3.9 years for the proteinuria endpoint to 5.5 years for chronic kidney disease. TDF-treated veterans took the drug for an average 1.3 years.
Multivariate analysis that considered demographics, HIV-related factors, other illnesses, and other antiretrovirals figured that each year of TDF therapy independently raised the risk of all three endpoints (Figure 2). The risk of chronic kidney disease jumped by one third for each year of TDF therapy.
Having chronic kidney disease, diabetes, or hypertension before starting TDF had little further impact on these kidney endpoints in people taking TDF. And stopping TDF during follow-up did not reverse these three signals of kidney toxicity (see note 17A).
The VA investigators suggested their findings "provide strong evidence that tenofovir may cause clinically significant toxicity to the kidney that is not reversible,"17 adding that "the balance between [TDF] efficacy and probable adverse effects requires further study." Further study is certainly in order for people taking TDF to prevent HIV rather than treat it. The researchers noted that their analysis is limited by their inability to measure eGFR directly, and no one can say whether the findings apply to women as well as men.
Two small single-center studies found, however, that impaired kidney function usually does improve when people stop TDF.18,19 The larger of these two studies, from Barcelona, found that TDF-associated kidney deficits reversed course in well over half of those who stop the drug.18 The study involved 183 people, 85% of them men, with a median age of 44 (interquartile range [IQR] 40 to 50). This group took TDF for a median of 39 months (IQR 22 to 63) and endured baleful changes in kidney markers during that time. The researchers used logistic regression to evaluate factors associated with a return to normal kidney markers, which they defined as eGFR at or above 60 mL/min, creatinine at or below 1.20 mg/dL, phosphate at or above 2.69 mg/dL, proteinuria below 30 mg/dL, and glycosuria below 20 mg/dL in people without diabetes.
A median of 22 months (IQR 13 to 49.5) after people stopped TDF, renal values returned to normal in 59% of this group, improved without reaching normal benchmarks in 10%, and did not improve in 31%. Median time to reaching normal values measured 4 months (IQR 2 to 15.75). Follow-up time stood below 12 months in 30% of people with no improvement in kidney values. Higher nadir CD4 count predicted higher odds of returning to normal kidney function (odds ratio [OR] 1.002 per 1-cell increase, P = 0.034), as did higher CD4 count when TDF stopped (OR 1.033 per 1-cell increase, P = 0.030). These associations, the researchers proposed, suggest "a role of preserved cellular immunity in renal recovery" after stopping TDF.
The second single-center study involved 80 people with TDF-associated kidney toxicity at an inner-city clinic in Chicago, including 86.5% with increasing serum creatinine and/or declining eGFR, 11% with proteinuria, and 2.5% with Fanconi syndrome.19 The investigators defined improved renal function as (1) eGFR rising to or above the baseline value, (2) qualitative or quantitative improvement in proteinuria, or (3) improvement or normalization of parameters defining Fanconi syndrome. Most study participants (84%) were men, and 74% were black. Median age stood at 53 years, and TDF duration averaged 27.4 months. Two thirds of the TDF regimens included a protease inhibitor (PI). Forty-nine people (61%) stopped TDF and 31 (39%) continued.
Kidney function values improved in 37 of 49 people (76%) who stopped TDF versus 17 of 31 (55%) who did not. Kidney disease progressed to end-stage renal disease in 2 of 49 people (4%) who stopped TDF. Multivariate analysis determined that stopping TDF nearly quadrupled odds of renal recovery (OR 3.76, 95% CI 1.26 to 11.27, P = 0.02). Factors that did not affect chances of recovery in this analysis were diabetes, hypertension, and use of other nephrotoxic drugs.
The VA investigators suggested that different mechanisms of TDF-induced kidney damage may explain why impairment improves in some people but not in others who stop TDF.17 TDF accumulation in the proximal renal tubule may be reversible, they proposed, whereas active tubular necrosis and tubulointerstitial scarring may not.
A large and long US multicenter study found a link between chronic kidney disease and regimens containing TDF plus a ritonavir-boosted PI, but TDF/nonnucleoside combos or a boosted PI without TDF did not send kidneys out of kilter.20 The study involved 3329 adults in HIV care who had at least one creatinine reading before and after starting antiretroviral therapy. One quarter (24.9%) took TDF with a ritonavir-boosted PI, 35.1% took TDF with a nonnucleoside, 11.9% took a PI without TDF, and 28.1% took a nonnucleoside without TDF. Median age stood at 40 years, 38.5% of study participants were black, and 18.7% were women. Rates of kidney disease risk factors were moderate to low -- 16.1% had hypertension, 14.9% had HCV infection, and 3.2% had diabetes. Median follow-up spanned 4.8 years, including a median of 23 weeks before antiretroviral therapy and 143 weeks on treatment.
Analyses adjusted for kidney risk factors determined that black race, HCV infection, and lower CD4 count and higher viral load during antiretroviral therapy swelled the risk of chronic kidney disease.20 Overall, antiretroviral therapy was associated with a slower eGFR decline. But taking a TDF/PI regimen (compared with a nonnucleoside without TDF) more than tripled the odds that eGFR would fall below 60 mL/min (OR 3.35, 95% CI 1.40 to 8.02, P = 0.006). Still, after 4 years taking a TDF/PI regimen, eGFR fell that low in only 5.7% of patients. Severe chronic kidney disease (eGFR below 30 mL/min) developed in only 16 people through more than 10,000 person-years of follow-up. Taking TDF without a boosted PI or taking a boosted PI without TDF did not independently hike the odds of an eGFR below 60 mL/min or 45 mL/min.
What about renal safety of TDF when combined with the integrase inhibitor raltegravir in previously untreated people? The STARTMRK trial randomized antiretroviral-naive people to raltegravir or efavirenz, both with TDF/FTC.21 After 3 years of followup, none of 281 people randomized to raltegravir and 1 of 279 randomized to efavirenz had creatinine levels at or above 1.9 times the upper limit of normal.
A meta-analysis of 17 studies comparing TDF regimens with non-TDF therapies found an average 3.92 mL/min lower creatinine clearance in TDF takers (95% CI 2.13 to 5.70) in 11 studies reporting that outcome.22 In 8 studies risk of acute renal failure was 0.7% higher in the TDF group (95% CI 0.2 to 1.2), but TDF takers did not have a higher rate of chronic kidney failure or end-stage kidney failure requiring long-term dialysis. Three of these studies had 86 to 96 weeks of followup, and 4 had 104 to 520 weeks. Renal function declined less in randomized trials than in observational studies, and less in people starting first-line TDF than in treatment-experienced people. This meta-analysis yielded no evidence that TDF heightened the risk of severe proteinuria or hypophosphatemia.
Among the longest individual studies of TDF in people with HIV are an international randomized trial comparing TDF/FTC with zidovudine/lamivudine (ZDV/3TC), each with efavirenz,23 an international randomized trial comparing TDF with stavudine in previously untreated adults,24 a EuroSIDA analysis of chronic kidney disease,25 and a study of HIV-positive people at two UK centers.26 (The meta-analysis22 included three of these studies.23,24,26) Side effect findings in the TDF trials must be interpreted with the understanding they excluded people with creatine clearance below 5023 or 6024 mL/min, people taking other kidney-toxic drugs,23 and people with a history of "clinically significant bone disease."23 As noted in the meta-analysis of 17 studies discussed above, renal function appears to drop less in randomized trials than in observational studies.22
The first trial involved 517 antiretroviral-naive adults, 14% of them women, with a median age of 37 years. Everyone had an eGFR above 50 mL/min when the trial began and creatinine below 1.5 mg/dL.23 After 144 weeks, more people stopped ZDV/3TC than TDF/FTC because of adverse events (11% versus 5%, P = 0.01), and no one dropped out because of kidney trouble. One person in the TDF group had a confirmed grade 1 creatinine elevation through 144 weeks, while 2 in the ZDV arm had a confirmed grade 2 elevation. Median eGFR dropped 12 mL/min (from 110 to 98 mL/min) in the TDF group while rising 1 mL/min (from 105 to 106 mL/min) in the ZDV group, a significant difference (P < 0.001).
The trial that established TDF efficacy in antiretroviral-naive people randomized 299 to TDF and 303 to stavudine, each with lamivudine and efavirenz.24 About 20% of participants were black. After 144 weeks of follow-up creatinine rose above 2 mg/dL in 2 of 296 randomized to TDF and 2 of 296 randomized to stavudine. Similar proportions in both arms had proteinuria (above 30 mg/dL) by week 144: 18% in the TDF arm and 23% in the stavudine arm. Creatinine clearance rose by an average 2 mL/min in the TDF group and 7 mL/min in the stavudine group through 144 weeks. No one dropped out of the study because of TDF-related kidney toxicity. A 1111-person 3-year analysis combining results of these two TDF trials23,24 determined that fewer than 1% in the TDF arms and the comparison arms had confirmed creatinine elevations above 1.5 mg/dL or serum phosphorus below 2 mg/dL.27
A EUROSIDA analysis of 6843 adults with at least three creatinine measures and a median follow-up of 3.7 years (IQR 2.8 to 5.7) found that TDF and three PIs -- atazanavir, indinavir, and lopinavir -- independently raised the risk of chronic kidney disease, defined as a confirmed eGFR falling below 60 mL/min or a 25% drop in eGFR for people starting below 60 mL/min.25 After adjustment for traditional risk factors, each year of TDF use raised the risk of chronic kidney disease about 15% (incidence rate ratio 1.16, 95% CI 1.06 to 1.25, P < 0.0001).
The UK cohort study involved 3439 HIV patients seen at two centers between January 1998 and December 2005.26 Follow-up ranged from 104 to 520 weeks. Among 843 people who took TDF during the study period, chronic kidney disease (eGFR below 60 mL/min for at least 3 months) developed in 26 people (3.1%), compared with an overall incidence of 2.4% in this study group. Of the 22 people who stopped TDF during follow-up, 21 did so because of kidney concerns. In people with chronic kidney disease, taking TDF was tied to a 3.7-fold faster decline in kidney function. And among people who started TDF during the study period, being 50 or older boosted the odds of chronic kidney disease 5.4 times, while an eGFR of 60 to 74 (versus 90 or higher) raised the odds 17.2 times.
Implications of Studies of Kidney Function in PrEP Candidates and TDF Takers
Only the bold would try to wrest overarching conclusions from the data payload explored in the preceding two sections. But a few suggestions seem feasible.
First, the limited studies of kidney function in HIVnegative people with a possibly high risk of infection offer no evidence that PrEP candidates, as a group, have notably fragile kidneys just waiting for a nephrotoxic insult.12,15 But in the MACS study of HIVpositive and negative gay and bisexual men, the negative group had concerning rates of two kidney risk factors -- hypertension (37%) and diabetes (13%).12 And although HIV-positive women in the HERS analysis had a significantly higher rate of renal lab abnormalities than HIV-negative women, more than 1 in 20 women in the negative group did have such warning signals.15 Findings like these buttress FDA7 and CDC28,29 advice to screen PrEP candidates for creatinine clearance, to probe for other kidney risk factors (including smoking) before prescribing TDF/FTC, and to continue monitoring kidney function in PrEP takers.
PrEP trials showing low rates of creatinine elevations or phosphorus slumps in HIV-negative people randomized to TDF offer some reassurance.4-6,11 Remember, though, that these trial participants got monitored more than PrEP takers would in clinical practice, and that follow-up in these trials ranged from 1 to 2 years.
Clinical trials and cohort studies of HIV-positive people taking TDF found low to modest rates of elevated creatinine, slowed creatinine clearance, or chronic kidney disease.23,27 But three large cohort studies in the United States and Europe linked TDF therapy to possibly ominous changes in kidney markers,17,20,25 though one of the US studies discerned a higher risk of chronic kidney disease with TDF only in people also taking a ritonavir-boosted PI.20
The study tying longer TDF use to three unpropitious kidney outcomes in US veterans raised the additional concern that these ill-trending renal markers did not improve readily when TDF stops.17,17A These investigators cite two other studies confirming incomplete reversibility of TDF-linked kidney changes, though one of these studies involved only 24 men30 and the other was cross-sectional.31 And two singlecenter studies documented improving kidney markers in most people with compromised kidney function who stopped TDF.18,19 Still, prudent prescribers will keep the veterans findings in mind when evaluating people for PrEP and charting their progress.
Bone Mineral Density in PrEP Users and Candidates
FDA regulators7 and CDC experts28,29 do not suggest screening all PrEP candidates for bone mineral density (BMD) before starting TDF/FTC PrEP. But, without differentiating between TDF/FTC for PrEP or HIV therapy, the FDA counsels prescribers to check BMD in people with "a history of pathologic bone fracture or other risk factors for osteoporosis or bone loss."7 Prescribing information suggests "supplementation with calcium and vitamin D ... may be beneficial" for people with a high risk of bone loss but notes that this strategy remains unstudied.
PrEP trials offer a look at BMD changes and fracture risk in HIV-negative people randomized to daily TDF, and the literature features several analyses of bone changes in HIV-negative people at risk of HIV infection or people with newly diagnosed HIV (Table 5). PrEP trial data on TDF-related bone toxicity must be interpreted cautiously because of the poor TDF/FTC adherence reported among many trial participants.
Table 5. Key Findings on Bone Health in HIV-Negative at-Risk People Taking or Not Taking TDF
BMD declined more with TDF than with placebo in the iPrEx trial,4,7 the TDF2 trial,6 and a TDF PrEP trial in gay men in San Francisco.32
In iPrEx 13% of participants randomized to TDF/FTC lost at least 5% of spine BMD.7
During 1 to 2 years of follow-up, no PrEP trials found a higher fracture rate with TDF than with placebo.4-7,9
Bone-related side effect rates in PrEP trials must be interpreted in the context of the relatively low adherence reported in these trials.
One in 10 gay/bisexual men in a TDF PrEP trial in San Francisco had low BMD when they entered the trial at a median age of 40 -- before they started taking TDF.32
Half of HIV-negative but at-risk men at a median age of 55 had osteopenia or osteoporosis in New York City.34
Seventeen of 33 men with primary HIV infection in the Netherlands and averaging 38 years in age had osteopenia or osteoporosis.35
Prevalence of low BMD was only 19% among HIV-negative at-risk women averaging 44.5 years of age in a New York City study.36
A WIHS cohort analysis of premenopausal women with and at risk of HIV logged similar BMD declines and similar low fracture incidence in the two groups through 2.5 years of follow-up.39
Another WIHS study recorded a similar fracture incidence in premenopausal women with and without HIV (1.8 and 1.4 per 100 person-years) through 5.4 years of follow-up.40
Methadone maintenance raised chances of low BMD in two studies of HIV-positive and negative women36,38 and in one study of HIV-positive and negative men.34
In the studies highlighted above, rates of low-BMD risk factors were high in both women and men at risk of HIV infection.
BMD, bone mineral density.
The iPrEx PrEP trial of HIV-negative men and transgender women who have sex with men4 included a bone substudy of 503 participants, summarized in the Truvada license.7 People randomized to TDF/FTC had greater declines in BMD than those randomized to placebo; these drops ranged from -0.4% to -1.0% across total hip, spine, femoral neck, and trochanter. Bone mineral changes migrated back toward baseline values after TDF/FTC stopped. While 6% of iPrEx participants randomized to placebo lost at least 5% of spine BMD during follow-up, 13% randomized to TDF/FTC lost that much. But fracture rates did not differ between the TDF/FTC group and the placebo group (1.7% and 1.4%), and BMD changes did not correlate with fractures.
In the TDF2 PrEP trial of high-risk Botswana men and women, people in the TDF/FTC group had significantly greater BMD drops in the forearm, hip, and lumbar spine than did people randomized to placebo.6 But fracture rates did not differ between the two groups (1.1% with TDF, 1% with placebo) through a median 1.1 years of follow-up. The Partners PrEP trial of TDF and TDF/FTC PrEP in HIVdiscordant African couples did not measure BMD. But again fracture rates during a median 1.9 years of follow-up were similar in people randomized to a TDF regimen and those randomized to placebo (0.8% and 0.6%).5,7 Fractures were even less frequent in the FEM-PrEP trial of African women, 0.1% in the TDF/FTC group and 0.2% in the placebo group.9
These placebo-controlled trials of TDF/FTC PrEP are unanimous in finding no excess fracture incidence with TDF/FTC PrEP versus placebo during 1 to 2 years of follow-up. But BMD did decline significantly more with TDF than with placebo in iPrEx4 and TDF2.6
A pre-iPrEx placebo-controlled trial of TDF PrEP in gay and bisexual men in San Francisco offers admonitory data on bone risk in US men who may take PrEP.32 Before anyone swallowed a single pill, DEXA scanning determined that 20 of 210 men (10%) had a BMD z score at or below -2.0 at the L2-L4 spine, total hip, or femoral neck. These men had a median age of about 40. Taking amphetamines inflated the odds of low BMD almost 6 times (OR 5.86, 95% CI 1.70 to 2.20), and using inhalants (poppers, amyl nitrate, nitrous oxide, glue) more than quadrupled the odds (OR 4.57, 95% CI 1.32 to 15.81). Men who took multivitamins, calcium, or vitamin D had lower chances of deficient BMD than did men not taking these supplements.
These researchers tracked BMD changes in 184 men, half of whom began TDF or placebo after a 9-month hiatus to appraise changes in pill-associated risk behavior.32 Compared with men who started placebo or no study drug, those starting TDF had a 1.1% net drop in BMD at the femoral neck (P = 0.004), a 0.8% decline at the total hip (P = 0.003), and a 0.7% dwindling at the L2-L4 spine (P = 0.11). After 24 months, 13% of men randomized to TDF versus 6% randomized to placebo or taking no study drugs had more than a 5% fall in BMD at the femoral neck, a nonsignificant difference (P = 0.13).
The investigators believe their findings "suggest that low BMD may pre-date HIV infection among men at risk for acquisition of HIV, and use of tenofovir in these individuals leads to a small but statistically significant decline in BMD."32
A MACS cohort analysis of HIV-positive and at-risk gay or bisexual US men at least 30 years old found a similar osteoporosis-related fracture incidence in the HIV-positive and negative groups.33 But after statistical adjustment for body mass index and race, men with HIV had a higher fracture incidence starting at age 50. This analysis involved 5106 men who made study visits every 6 months at some point between 1996 and 2011. Age averaged 45.2 in men with HIV and 47.5 in men without HIV, and respective proportions of whites were 70% and 82%. In the HIV and no-HIV groups, 31% and 24% smoked and 42% and 45% smoked at the past.
During follow-up the MACS team counted 53 FRAXdefined fractures in the HIV group and 50 in the HIV-negative groups for crude incidence rates of 0.15 per 100 person-years with HIV and 0.13 per 100 person-years without HIV. But after statistical adjustment for body mass index and race, incidence was higher in HIV-positive men than negative men 50 to 64 years old (1.78 versus 0.74 compared with 30- to 39-year-olds) and higher in HIV-positive men than negative men 65 or older (3.50 versus 2.44 compared with 30- to 39-year-olds) (Figure 3).
A New York City study compared BMD in men with HIV and at risk of HIV, all of them at least 49 years old (median 55).34 This study involved 328 HIV-positive men in the Cohort of HIV at-risk Men's Prospective Study (CHAMPS) and 231 HIV-negative cohort members. HIV-negative men had a high risk of HIV infection because they injected drugs or had high-risk sex. Most men (89%) used illicit drugs, and 56% were black. High proportions of HIV-positive and negative men had classic risk factors for low BMD, including 90% who smoked or once smoked, 47% in a methadone maintenance program, 30% who used heroin in the past 5 years, 52% who exercised less than once a week, and 54% with serum testosterone below 300 ng/dL.
Of the 559 men studied, 299 (54%) had osteopenia or osteoporosis, and that rate did not differ significantly between men with and without HIV (55% versus 51%, P = 0.4).34 The osteopenia rates in these men were similar to national estimates among white men 50 and older. But the proportion of men with osteoporosis in this study, 14%, was higher than in the general population.
Statistical analysis adjusted for age, weight, race, testosterone level, and prednisone and illicit drug use found lower BMD at the femoral neck and lumbar spine in men with than without HIV. HIV infection, older age, nonblack race, lower weight, low testosterone, prednisone use, heroin use, and current methadone maintenance were independently associated with low BMD at the femoral neck, lumbar spine, or both sites. Osteopenia or osteoporosis independently raised the fracture risk, but HIV infection did not.
A Netherlands study of 33 young men just infected with HIV found that more than half had osteopenia or osteoporosis.35 Some evidence from this study hinted HIV itself accounted for low BMD in at least some of these men, but other evidence suggested low BMD preceded HIV infection in these men.
Of the 33 men assessed, 30 were gay or bisexual, none were injection drug users, and none had HCV or HBV infection.35 Their age averaged 38 years, and their body mass index averaged 22.7 kg/m2 (within the normal range of 18.5 to 24.9). Only 1 man had taken antiretrovirals -- for 9 days -- when DEXA scans measured BMD of the lumbar spine, femoral neck, and total hip. Low BMD risk factors abounded in these men: Twenty-six men (79%) were white, 22 (67%) currently used drugs, 18 (55%) smoked, and 7 (21%) downed more than 3 alcoholic drinks a day. Ten men (30%) had low osteocalcin levels, which may betoken flagging bone formation. Twelve men (36%) had broken a bone before getting infected with HIV. About half of these men also had some bone-boosting habits -- multivitamin use by 17 men (52%) and strenuous exercise for at least 20 minutes at least 3 times a week by 19 men (58%).
Average lumbar spine t score (-0.8) and z score (-0.7) and average femoral neck t score (-0.5) were significantly lower than in a reference population. Fifteen of 33 men (45%) met World Health Organization criteria for osteopenia, and 2 of 33 (6%) met osteoporosis criteria. "These numbers," the authors observed, "are much higher than would be expected in a relatively young male population like ours."35
Can recent HIV infection -- and the resulting lofty viral load -- explain low bone density in these men? Or did their multiple bone risk factors, maybe coupled with genetics, account for these formidable rates of osteopenia and osteoporosis? Perhaps both. But it's hard to dismiss the big list of risk factors -- including an impressive fracture history -- these men brought into the study. One statistical analysis these researchers ran suggested spiking viremia during primary HIV infection did contribute to low BMD in these men: Linear regression analysis adjusted for age and body mass index saw a link between higher viral load and lower total hip t score (β -0.2, P = 0.02). But reduced BMD in these men could not be tied to biochemical evidence of brisk bone turnover or systemic inflammation, which the researchers noted might be expected during primary HIV infection. Citing the San Francisco study of HIV-negative gay and bisexual men,32 the Dutch team proposed their findings also raise the question "whether it is actually the recent HIV-1 infection causing rapid bone loss shortly after transmission, or whether these bone disorders predate HIV infection and are caused by other risk factors."35
Several studies scrutinized bone variables in HIVnegative women with some risk of HIV infection. A 2001-2003 study of HIV-positive women and negative women at risk of infection found that HIV independently raised the risk of low BMD, but only in nonblack women.36 Overall prevalence of low BMD in these middle-aged women (27% in the HIV group and 19% in the negative group) lagged national estimates of osteopenia and osteoporosis, perhaps because of the high proportions of black and overweight women in this cohort.
The study involved 263 women with HIV and 232 without HIV in the New York City-based Menopause Study, a longitudinal analysis of menopause and its impact on women with and at risk of HIV infection.36 The women averaged 44.5 years in age; racial/ethnic proportions were 59% and 44% black in the HIV and no-HIV groups, 34% and 42% Hispanic, and 6% and 12% white. Only 8% of women were postmenopausal, though about 74% were rated perimenopausal (older than 40 and not amenorrheic). Proportions of overweight women (25 to 29.9 kg/m2) were 36% with HIV and 22% without HIV, and respective proportions of obese women were 32% and 62%.
Moderately high proportions of these women -- 61% with HIV and 54% without HIV -- reported regular exercise, though a similar proportion -- 53% in both groups -- reported watching more than 4 hours of TV daily. High proportions (63% with HIV and 72% without HIV) smoked, while 27% and 19% had smoked in the past. About 8% of women had used prednisone, and about 15% had used estrogen. About 20% had an earlier fracture. Drug use was not rare in women with or without HIV: heroin in past 5 years (22% and 33%), cocaine in past 5 years (44% and 45%), and current methadone maintenance (23% and 44%).
Compared with HIV-negative women, those with HIV had significantly lower BMD at the femoral neck and lumbar spine, significantly lower t scores at both sites, and significantly lower z scores at the femoral neck. But low BMD affected these women less often than women in the general population. While 51% to 70% of white women in the US at least 50 years old have reduced BMD (up to 50% with osteopenia and 20% with osteoporosis),37 nonblack women in this study group had a 25% prevalence of osteopenia and a 5% rate of osteoporosis. The researchers suggested the low overall rates of reduced BMD in their cohort (27% of women with HIV and 19% without HIV) reflect the high proportion of black women (who have a diminished risk of low BMD compared with whites) and the "extremely high prevalence" of overweight and obesity.
Linear regression analysis determined that HIV independently raised the risk of low BMD (defined as a t score at least 1 standard deviation below the average peak bone mass in young adult women) at the femoral neck and lumbar spine (β -0.026, P = 0.02; β -0.041, P < 0.01). Classic independent risk factors were older age, nonblack race, low weight, prednisone use, cocaine use, prior fracture, and methadone maintenance. In race-stratified analyses, HIV was independently associated with low femoral neck BMD (β -0.04, P < 0.01) in nonblack women, but the HIV association did not hold in black women. In women with HIV, neither nucleosides nor PIs were associated with low BMD.
This New York Menopause Study in women36 and the New York CHAMPS study in HIV-positive and at-risk men34 both linked methadone maintenance to low BMD. A recent comparison of BMD in HIVnegative women on methadone and healthy controls confirmed a lower total hip BMD in the methadone group, but not lower lumbar spine or femoral neck BMD.38 The study involved 11 young women taking methadone after heroin addiction and 30 healthy women without a heroin problem and not taking methadone. Ages ranged from 20 to 29 in the two groups, and the methadone group had taken this agent for 1.5 to 9 years (median 3). No women in either group had former or current low weight, though average body mass indices leaned toward the lower end of normal (21.9 kg/m2 with methadone and 20.5 kg/m2 without). Women taking methadone had other bone risk factors, including smoking, alcohol use, and cocaine use. DEXA scans found equivalent BMD in the spine and femoral neck in the methadone group and the comparison group, and marginally lower total hip BMD in women on methadone maintenance (P = 0.054 for BMD, P = 0.049 for t score). The researchers proposed that "long-term methadone substitution in HIV-negative women seems to slightly affect bone mass density."38
Two analyses of BMD and fractures in premenopausal women with HIV and at risk of HIV infection found lower BMD in the HIV group but no difference in fracture rates.39,40 Both studies come from the Women's Interagency HIV Study (WIHS), which recruits HIV-positive and at-risk women in the Bronx, Brooklyn, Washington, DC, Los Angeles, San Francisco, and Chicago.
The first study involved 100 women with HIV and 68 uninfected women who had DEXA scans of the femoral neck and lumbar spine at visits separated by a median of 2.5 years.39 At the first visit women with HIV had 5% lower BMD at both sites, but the annual drop in BMD through follow-up did not differ between groups. Statistical analysis adjusted for age, weight, and BMD at the initial visit confirmed similar BMD declines in women with and without HIV. Self-reported fracture incidence was nonsignificantly higher in the HIV-negative group (1.03 versus 0.74 per 100 person-years without and with HIV, P = 1.0).
As in other studies of US women at risk of HIV infection, this WIHS contingent carried more than a few bone risk factors: 65% smoked at the first visit and 78% ever smoked, 56% drank alcohol, 46% ever used cocaine, 23% injected drugs, 31% tested positive for HCV, and 5% had diabetes.39 On the plus side -- as far as BMD is concerned -- body mass index averaged 30.2 kg/m2, in the obese range. And 26% of women took vitamin D. Low weight and alcohol use were associated with low BMD in the whole study group. In HIV-positive women, CD4 count and antiretroviral class did not predict declining BMD.
A larger and longer comparison of mostly premenopausal HIV-positive and negative WIHS women found a fracture incidence of 1.8 per 100 person-years in women with HIV and 1.4 in women without HIV, a nonsignificant difference (P = 0.18).40 Median followup measured 5.4 years. Multivariate analysis did not tease out an association between HIV infection and fracture in this study. And HIV-positive women had a bone-risk disadvantage compared with HIV-negative women because they were older (average 40 versus 36 years, P < 0.0001), weighed less (74.5 versus 79.7 kg, P < 0.001), and were more likely to be postmenopausal and to have HCV coinfection. Hip and wrist fracture incidence rates in the HIV-positive women (0.2 and 0.3 per 100 person-years) were similar to hip and wrist fracture rates in the general population of premenopausal women in the United Kingdom.41,42
As in the smaller WIHS study,39 HIV-negative women in the larger analysis40 had their share of low-BMD risk factors: 51% smoked, 21% rated themselves moderate or heavy drinkers, 19% had a prior fracture, 14.5% had HCV infection. In the plus column, 28% took vitamin D. Bivariate analysis of the entire study group linked an array of classic risk factors to incident fracture: older age, white race, self-reported menopause, prior fracture, HCV infection, higher diastolic blood pressure, cigarette smoking, and injection drug or opiate use.
Bone Changes With Long-Term TDF Therapy
Two longitudinal studies and three randomized trials implicate TDF in BMD loss in HIV-positive men and women (Table 6). A large Veterans Affairs (VA) analysis determined that every year of TDF use inflates the risk of osteoporotic fracture 12%.43 The longitudinal studies and trials did not tally fracture rates or found no higher rate with TDF than with other regimens.
Table 6. Key Findings on Bone Health in HIV-Positive People on Long-Term TDF
A prospective study of 33,439 US veterans determined that every year of TDF use boosts the risk of osteoporotic fracture 12%.43
Longitudinal analysis in a US Nutrition for Healthy Living Study involving 283 men and 96 women linked TDF use to an average 2.04% drop in total BMD among men and an average 1.74% drop in premenopausal women.44
Longitudinal analysis of 483 men and 188 women in Spain determined that taking TDF at the time of the most recent DEXA scan raised odds of declining BMD more than 40%.45
After 3 years of follow-up in a 517-person trial comparing TDF/FTC with ZDV/3TC in previously untreated adults, fracture rates were low and similar in the two treatment arms.23
After 3 years of follow-up in a 602-person trial comparing TDF/3TC with stavudine/3TC in previously untreated adults, BMD waned more at the lumbar spine and hip with TDF/3TC, but fewer people in the TDF group broke a bone.24
After 96 weeks of follow-up in a trial that randomized antiretroviral-naive adults to TDF/FTC or ABC/3TC, BMD declined more at the spine and hip in the TDF group.46 Fracture rates did not differ between the two arms.
BMD, bone mineral density.
The VA study involved 56,660 US veterans seen from 1998 through 2009, 98% of them men.43 Two thirds took antiretrovirals for at least 1 month. During follow-up 951 vets had an osteoporotic fracture, ascertained by ICD-9 code and defined as the first new spine, hip, or wrist fracture "selected on the basis of their likelihood of being related to osteoporosis." Veterans who sustained fractures were generally middleaged and only moderately older than those who did not (46 versus 44 years). People with fractures were more likely to be white (57% versus 45%) and more likely to smoke (56% versus 32%), have diabetes (25% versus 15%), have a body mass index below 20 kg/m2 (49% versus 33%), and have HCV infection (51% versus 31%) (P < 0.0001 for all comparisons).
The VA investigators sized up fracture risk factors in two multivariate models; the first factored in race, age, tobacco use, diabetes, body mass index, chronic kidney disease, HCV status, and cumulative exposure to TDF, abacavir, zidovudine or stavudine, any boosted PI, and any nonnucleoside; the second model added concomitant exposure to other antiretrovirals. Neither model considered other possibly telling variables, such as exercise, vitamin D or calcium, or use of steroids, alcohol, or illicit drugs; and the researchers could not evaluate BMD. Because the cohort included few women, results may not apply to them.
Through an average follow-up of 5.4 years, both multivariate models figured that every year taking TDF hiked the osteoporotic fracture risk 6%, but in both analyses the 95% confidence interval just crossed 1.0 (0.99 to 1.12 in model 1 and 0.99 to 1.14 in model 2).43 When the researchers limited the analyses to 33,439 vets who entered the cohort in the combination antiretroviral era (starting January 1, 1996), every year taking TDF independently upped the fracture risk about 12% (model 1 hazard ratio [HR] 1.13, 95% CI 1.05 to 1.21, P = 0.001; model 2 HR 1.12, 95% CI 1.03 to 1.21, P = 0.011). Adding a boosted PI to a TDF regimen inflated the risk slightly more (HR 1.16, 95% CI 1.04 to 1.30). Cumulative antiretroviral use did not make fractures more likely, but several classic risk factors did: white race, older age, tobacco use, and body mass index below 20 kg/m2.
Longitudinal analysis of total body BMD in HIV-positive men and women in the Nutrition for Healthy Living Study linked greater bone loss to steroids and two antiretrovirals -- TDF and didanosine.44 The Nutrition for Healthy Living Study is a prospective cohort of HIV-positive adults living in Massachusetts and Rhode Island. This analysis included 283 men and 96 women seen between August 1996 and September 2003 who had at least two whole-body DEXA scans at least 1 year apart. Median age was 42.7 in men and 39.4 in women; 59.4% and 34.4% were white, 25.4% and 51.0% were black. Smoking rates were high in both men and women (43.3% and 66.7%), as was a history of injection drug use (32.2% and 44.8%). Among women, 17.7% were postmenopausal. While 30% of men reported strength training in the previous week, only 11.6% of women did.
Statistical analysis adjusted for age, race, sex, menopause, and smoking linked greater loss of total BMD to TDF use, longer didanosine use, prednisone or hydrocortisone use, lower body mass index, and lower albumin.44 TDF use conferred an average 2.04% drop in total BMD among men and an average 1.74% decline in premenopausal women. Strength training mitigated loss of total BMD.
A longitudinal study of 671 antiretroviral-treated people in Spain included 483 men (72%) and 188 women, only 18 of them (10%) postmenopausal.45 Median age for the whole group was 42.1 years, median time on antiretroviral therapy 7.4 years, and median time on tenofovir 2.2 years. Almost half of the study group (47.5%) had osteopenia, and almost one quarter (23%) had osteoporosis.
Among people who took TDF for 1 year or less, 20% had osteoporosis, while in those who took TDF for more than 5 years, 37% had osteoporosis.45 Taking TDF at the time of the most recent DEXA scan upped the odds of declining BMD 44% (OR 1.44, 95% CI 1.03 to 2.20, P = 0.03). Among 105 people with at least 5 years of follow-up, DEXA scans confirmed progression to osteopenia in 18% and to osteoporosis in 29%. Odds of BMD loss or progression to osteopenia or osteoporosis rose with longer time taking TDF (OR 1.08, 95% CI 1.03 to 1.14, P < 0.0019), longer time taking a PI (OR 1.18, 95% CI 1.12 to 1.24, P < 0.0001), and current PI use (OR 1.64, 95% CI 1.35 to 2.04, P < 0.0001)
Through 144 weeks of follow-up in the 517-person international trial comparing TDF/FTC with ZDV/3TC (both with efavirenz) in antiretroviral-naive adults, 6 people in the TDF group and 8 in the ZDV/3TC group broke a bone.23 Trauma caused all fractures, and the investigators attributed none of the breaks to study drugs. The researchers did not report changes in BMD. This trial excluded people with a history of "clinically significant bone disease."
After 144 weeks in the trial comparing TDF/3TC with stavudine/3TC in 602 previously untreated adults, researchers charted a greater drop from baseline lumbar spine BMD in the TDF group (-2.2% TDF versus -1.0% stavudine, P = 0.001).24 BMD faded even more at the hip in both study groups, and the difference between groups approached statistical significance (-2.8% TDF and -2.4% stavudine, P = 0.06). The researchers noted, though, that BMD waning generally occurred through weeks 24 to 48 then stabilized. Five people randomized to TDF and 11 randomized to stavudine broke a bone during follow-up.
ACTG protocol A5224s was a substudy of a trial that randomized antiretroviral-naive adults to TDF/FTC or abacavir (ABC)/3TC with open-label efavirenz or atazanavir/ritonavir.46 Of the 269 study participants, 229 (85%) were men, and median age was similar across the four arms (38 years overall). Almost half of enrollees (47%) were white non-Hispanic, 33% were black non-Hispanic, and 16% were Hispanic. Almost one third (32%) broke a bone in the past.
After 96 weeks spine BMD decreased significantly more in the TDF/FTC arms than in the ABC/3TC arms (-3.3% versus -1.3%, P = 0.004), as did hip BMD (-4.0% versus -2.6%, P = 0.024).46 ABC/3TC plus efavirenz was the only combination not linked to a significant 96-week drop in spine BMD. From study entry to week 48, declines in BMD were greater with TDF/FTC than with ABC/3TC at the lumbar spine (-1.66%, P = 0.005) and hip (-1.43%, P = 0.007). But from weeks 48 through 192, mean percent change in BMD per year did not differ between the two groups. Regression analysis that factored in age, sex, race/ethnicity, and pretreatment viral load, CD4 count, and body mass index reckoned significant associations between ABC/3TC (versus TDF/FTC) and greater BMD at both the spine (parameter estimate 1.90, P = 0.003) and hip (parameter estimate 1.28, P = 0.033) at week 96. About 1 in 20 people had a trauma-related fracture during 96 weeks of followup, but neither fracture rate nor time to first fracture differed by treatment assignment.
Implications of Studies of BMD and Fractures in PrEP Candidates and TDF Takers
What can one make of the bone data medley from studies of people who run some risk of HIV infection and may consider TDF/FTC PrEP? First, the PrEP trials themselves show that taking TDF to ward off infection depletes bone mineral density -- but only in small proportions of people during these trials' 1 to 2 years of follow-up.4,6,7,32 And taking TDF for PrEP had no impact on fracture risk during any of these placebo-controlled trials.4-7,9,32
How longer-term TDF PrEP will affect bone health depends on how consistently people take their PrEP pills and what other bone risk factors they have. Several studies of HIV-negative but at-risk men and women in the United States and the Netherlands leave no doubt that people likely to consider PrEP bear a hefty burden of low-BMD risk factors,32-36,38-40 including cigarette smoking, use of alcohol and injected or noninjected drugs, methadone maintenance, lack of exercise, previous fractures, diabetes, and HCV infection.
A study of HIV-negative gay and bisexual men recruited for a TDF PrEP trial in San Francisco found that 10% had low BMD of the spine, total hip, or femoral neck before they started PrEP, and that taking amphetamines or inhalants boosted the risk of low BMD.32 This finding resonated in a Dutch study of men (91% gay or bisexual) with primary HIV infection, 45% of whom had osteopenia and 6% of whom had osteoporosis.35 The researchers found mixed evidence on whether the spiking viremia of acute infection caused or contributed to this high rate of depleted BMD or whether that rate could be more firmly tied to pre-HIV risk factors. Rates of osteopenia or osteoporosis were high and similar in 49-year-old and older HIV-positive and negative but at-risk men in a New York City study (55% and 51%).34 A large majority of men in this study, 89%, used illicit drugs, and 47% were on methadone maintenance.
Together these findings indicate that middle-aged male PrEP candidates in the United States and perhaps Western Europe -- including gay men and drug injectors -- may have compromised bone health before starting TDF/FTC PrEP. Clinicians considering PrEP for men like these would do well to heed FDA advice to check them for "a history of pathologic bone fracture or other risk factors for osteoporosis or bone loss."7 A pre-PrEP bone scan may be in order for some men. Notably, the early San Francisco PrEP trial found that men taking multivitamins, calcium, or vitamin D trimmed their risk of low BMD.32 And the FDA suggests calcium or vitamin D supplements may have a role in slowing BMD decline.7
Unlike these studies of men at risk of HIV infection or with acute HIV infection,32-35 studies of at-risk mostly premenopausal US women in two cohorts did not find an undue burden of osteopenia or osteoporosis.36,38-40 Two studies saw links between methadone maintenance and low BMD in women36,38 (as did one in HIV-positive and at-risk men34). All these studies recorded lofty rates of classic bone risk factors in women with and without HIV, and these risk factors inflated the odds of low BMD in statistical analyses. Providers discussing PrEP with women should review the list of bone risk variables in Table 1 and might consider DEXA scanning, vitamin D, and calcium for women with an apparently high risk.
Three longitudinal studies of HIV-positive people were unanimous in linking TDF use to either a higher osteoporotic fracture rate43 or to dwindling BMD.44,45 The large Veterans Affairs study that established a magnified fracture risk with TDF is limited in that it could not determine changes in BMD; it could not clinically confirm fractures, which were ascertained by ICD-9 codes; and almost all study participants were men.43 The two much smaller longitudinal studies that confirmed declining BMD in people taking TDF did not report fracture rates.44,45 A Spanish study tied TDF use to progression from normal BMD to osteopenia or osteoporosis through 5 years of follow-up.45 Three randomized trials found equivalent or lower fracture rates with TDF regimens than with comparison regimens.23,24,46 But two of these studies confirmed greater declines in BMD with TDF combinations than with non-TDF combinations.24,46 In the two trials that tracked BMD, the dips occurred mostly in the first year of therapy and then stabilized.
Bottom Line: Low TDF Toxicity Risk, but Caution Advised
Anyone who prescribes antiretrovirals or scans FDA prescribing information for TDF7 knows that this reverse transcriptase inhibitor can muddle kidney function or deplete bone mineral. Gilead Sciences, TDF's maker, plainly acknowledges the drug's toxic potential in its full-tilt development of GS-7340, a defanged TDF facsimile the company hopes will stymie HIV better than TDF but with less toxic sting.47-50
Although two large randomized trials left no doubt that TDF has a cleaner safety record than the nucleosides it displaced, zidovudine and stavudine,23,24,27 long-term TDF therapy clearly poses some kidney and bone risk. Some research indicates that TDFlinked kidney toxicity dissipates when TDF stops,18,19 but a 10,000-person Veterans Affairs study found that it may not.17 Another prospective veterans study figured that every year of TDF use boosts the risk of osteoporotic fracture 12%,43 while other cohort studies and trials confirmed dwindling bone mineral density with TDF but found no greater fracture risk.23,24,44,46 TDF PrEP trials showed that the drug leaves a thin toxic trail in kidney and bone of HIV-negative people, though toxicity rates were low during the 1 to 2 years of follow-up in these studies.4-6,9,32
Cohort studies leave no doubt that men and women at risk of HIV infection have histories full of kidney and bone risk factors, summarized in Table 1. And some -- but hardly all -- studies of HIV-negative people disclosed an above-average rate of kidney and bone disease in these people.
PrEP prescribers should keep all this in mind when pondering risks and benefits of TDF/FTC with PrEP candidates. And they should follow FDA and CDC advice to avoid PrEP in people with creatinine clearance below 60 mL/min and to consider DEXA scans for candidates with a past fracture or other bone loss risk factors.7
Brooks RA, Landovitz RJ, Kaplan RL, Lieber E, Lee SJ, Barkley TW. Sexual risk behaviors and acceptability of HIV pre-exposure prophylaxis among HIV-negative gay and bisexual men in serodiscordant relationships: a mixed methods study. AIDS Patient Care STDS. 2012;26:87-94.
Peterson L, Taylor D, Roddy R, et al. Tenofovir disoproxil fumarate for prevention of HIV infection in women: a phase 2, double-blind, randomized, placebo-controlled trial. PLoS Clin Trials. 2007;2:e27.
17A. "Among those who discontinued tenofovir use, the time period following cessation was not significantly associated with either higher or lower risks of proteinuria (HR=1.05 per year, 95% CI: 0.93-1.18, p = 0.41) or rapid decline (HR=1.05 per year, 95% CI: 0.94-1.16, p=0.42), although there was a marginal association of time off tenofovir with [chronic kidney disease] (HR=1.22 per year, 95% CI: 0.99-1.50, p=0.055). All hazard ratios remained greater than unity, which suggests that the effects of tenofovir on kidney disease risk were not reversible following discontinuation."
Bonjoch A, Echeverría P, Perez-Alvarez N, et al. High rate of reversibility of renal damage in a cohort of HIV-infected patients receiving tenofovir-containing antiretroviral therapy. Antiviral Res. 2012;96:65-69.
Arribas JR, Pozniak AL, Gallant JE, et al. Tenofovir disoproxil fumarate, emtricitabine, and efavirenz compared with zidovudine/lamivudine and efavirenz in treatment-naive patients: 144-week analysis. J Acquir Immune Defic Syndr. 2008;47:74-78.
Gallant JE, Winston JA, DeJesus E, et al. The 3-year renal safety of a tenofovir disoproxil fumarate vs. a thymidine analogue-containing regimen in antiretroviral-naive patients. AIDS. 2008;22:2155-2163.
Jacobson DL, Spiegelman D, Knox TK, Wilson IB. Evolution and predictors of change in total bone mineral density over time in HIV-infected men and women in the Nutrition for Healthy Living Study. J Acquir Immune Defic Syndr. 2008; 49:298-308.
Bonjoch A, Figueras M, Estany C, et al. High prevalence of and progression to low bone mineral density in HIV-infected patients: a longitudinal cohort study. AIDS. 2010;24:2827-2833.
Callebaut C, Margot N, Stepan G, Tian T, Miller M. Virological profiling of GS-7340, a next-generation tenofovir prodrug with superior potency over TDF. 52nd Interscience Conference on Antimicrobials and Chemotherapy. September 9-12, 2012. San Francisco. Abstract H-552.
Ruane P, DeJesus E, D Berger D, et al. GS-7340 25 mg and 40 mg demonstrate superior efficacy to tenofovir disoproxil fumarate 300 mg in a 10-day monotherapy study of HIV-1+ patients. 19th Conference on Retroviruses and Opportunistic Infections. March 5-8, 2012. Seattle. Abstract 103.
Markowitz M, Zolopa A, Ruane P, et al. GS-7340 demonstrates greater declines in HIV-1 RNA than TDF during 14 days of monotherapy in HIV-1-infected subjects. 18th Conference on Retroviruses and Opportunistic Infections. February 27-March 2, 2011. Boston. Abstract 152LB.
Babusis D, Phan TK, Lee WA, Watkins WJ, Ray AS. Mechanism for effective lymphoid cell and tissue loading following oral administration of nucleotide prodrug GS-7340. Mol Pharm. 2012 Jul 12. Epub ahead of print.