IAS: Screening for B*5701 reduces abacavir hypersensitivity incidence to zero

Testing for the genetic variant that can cause the hypersensitivity reaction to the nucleoside (NRTI) drug abacavir (Ziagen) reduced the incidence of ‘immunologically-confirmed’ hypersensitivity reactions to zero in a large randomised controlled trial, the 4^th IAS Conference heard during the first of several late-breaker sessions on Wednesday.

Results from another study in African-Americans appear to suggest that genetic testing should be routine for people of all ethnicities.

PREDICTing hypersensitivity to abacavir

Abacavir – also found in the co-formulations Kivexa and Trizivir – can cause a hypersensitivity reaction (HSR) in a minority of people. A strong association between HSR and the B*5701 allele (genetic variant) has already been established, and many HIV clinics are already testing for it, particularly in the United Kingdom.

PREDICT – a large randomised study, funded by the drug’s manufacturer, GlaxoSmithKline (GSK) – was set up to establish whether the presence of B*5701 was a necessary condition for the reaction, and whether withholding abacavir from people with B*5701 would eliminate ‘true’ HSRs. The PREDICT study appears to show that this is the case.

The study recruited 1956 patients in eight countries (Australia, Italy, France, Germany, Romania, Russia Spain and the UK). All were tested for B*5701 but were randomised into a control arm who all started abacavir-containing HAART (the test results being blinded until the end of the study) and an intervention arm where abacavir was withheld from B*5701-positive patients. Patients’ own physicians were blinded as to whether their patients were B*5701 positive in either arm, although the study was not placebo-controlled. Consequently, physicians knew if their patients were not taking abacavir, but not why.

If, in the opinion of their physician, patients in either arm developed symptoms consistent with an HSR (i.e. ‘clinically-suspected’), they were then given a patch test. This involves exposing a small area of skin to different concentrations of abacavir; redness and swelling, sometimes with blistering, indicates a positive result and this was regarded as an ‘immunologically-confirmed’ HSR. About 20% of patients who were tolerant to abacavir were given a patch test too: none had a positive result.

After drop-outs, 847 patients in the control arm and 842 in the intervention arm were evaluable for immunologically-confirmed hypersensitivity reactions. Most patients (84%) were of white ethnicity and 12% were of African ethnicity or descent.

Suspected versus confirmed hypersensitivity

Although there was an equal proportion of patients with B*5701 in both the intervention arm and control arm (5.7% vs. 5.8%) fewer than half of the patients in intervention arm had clinically-suspected HSRs (3.4% vs. 7.8%; OR 0.4; p <0.0001).

After patch testing, the difference in the proportion of patients with immunologically-confirmed, i.e. ‘true’ HSRs was significant between the two arms: 2.7% in the control arm and zero in the intervention arm (OR 0.03; p < 0.0001). In other words, no-one who had a suspected HSR but was negative for B*5701 had a positive patch test.

In the control arm, on the other hand, fewer than half of symptoms diagnosed as possible HSRs were in fact hypersensitive: just 30 out of 66 (45.5%) patients with clinically-suspected HSR had it immunologically confirmed.

The negative predictive value of the B*5701 test, i.e. its ability to predict that patients would not experience an HSR if they were negative for B*5701, was 100% in the intervention arm (positive = no abacavir) and 95.5% in the control arm (all patients given abacavir).

White versus non-white patients were 2.19 times as likely to experience a clinically-suspected HSR (p = 0.0242) and 4.21 times as likely to have an immunologically-confirmed one, although this was not statistically significant (p = 0.1139), (See more below on the SHAPE Study for B*5701 in African-Americans.)

All immunologically-confirmed HSRs had multiple symptoms: about two-thirds had the presence of at least three symptoms (fever, rash plus another systemic symptom) and a third consisted of fever plus rash alone. In contrast, a high proportion of suspected HSRs that were not immunologically-confirmed tended to consist of one of more constitutional symptoms such as diarrhoea or respiratory symptoms and lack one or both of the defining symptoms of fever or rash.

Apart from ethnicity, the only other factors strongly associated with a clinically-suspected HSR were either existing protease inhibitor therapy (OR 1.86; p = 0.0094) or starting therapy with an NNRTI at the same time as abacavir (OR 3.19; p = 0.0011). In both cases, and especially with new NNRTIs, it is easy to see why doctors might mistake a reaction to one of the other drugs in the regimen for an abacavir HSR.

A secondary endpoint of the study was the unscheduled use of healthcare resources. Seven patients in the intervention arm versus 24 in the non-intervention arm spent time as in-patients in hospital, with one control patient being admitted to intensive care. Ninety-two patients in the intervention arm had unscheduled outpatient visits compared with 118 in the control arm.

Predicting HSR in African-Americans

Later in the day, the conference also heard the results of the SHAPE trial, a retrospective case-control study of abacavir HSRs in African-Americans. The controls consisted of 206 African-Americans and 202 individuals of white ethnicity, who were enrolled in other GSK studies and had taken (and tolerated) abacavir. Blood samples were tested, largely retrospectively, for B*5701. The prevalence of B*5701 in this abacavir-tolerant group was 4% in whites and 1% in blacks which, given that 55% of patients with B*5701 do not get an HSR, indicates overall prevalences of about 7.2% and 1.8% respectively.

The cases were 130 white and 69 African-American patients who developed a suspected HSR and were given a confirmatory patch test. All patients with a positive patch test subsequently tested B*5701 positive. However, while 44% of whites with suspected HSRs were later confirmed as B*5701 positive, only 14% of African-Americans were. The proportion who tested positive with a patch test was even lower: while around one-third of white patients with suspected HSRs had a positive skin patch test, fewer than 8% of African-American test positive. This may be partly because inexperienced doctors find it more difficult to evaluate skin reactions in black skin.

Almost all (96%) of suspected HSRs that were skin-patch positive involved fever compared with only 68% of patch test negative suspected HSRs. Ninety-one per cent of skin-patch positive suspected HSRs involved at least three symptoms versus 63% on patch-negative suspected HSRs. All but one skin-patch positive HSRs occurred within 20 days of initiation of abacavir therapy, while negative ‘HSRs’ were often detected weeks later. The exception was on confirmed HSR at 30 days: it was speculated that the patient might have delayed taking their abacavir after it was prescribed.

Although this study is not as rigorous at the PREDICT study, it confirms that suspected HSRs are more likely to prove to be ‘false positives’ in African American patients than in those of white ethnicity and more often lack the defining symptoms of fever and/or rash.

Dr Elizabeth Philips, presenting, concluded that standardised B*5701 testing in patients of all ethnicities will therefore reduce instances of misdiagnosis of HSR and also situations in which abacavir is inappropriately withdrawn from patients who could have benefited from it.

Clinical implications and recommendations

Although just over half of patients who are positive for B*5701 do not develop an HSR, it is thought that they are at very high risk of developing one, and GSK recommends that no patient positive for B*5701 is given abacavir.

GSK is also concerned that physicians do not use patch testing as a substitute for B*5701 screening; it is purely a confirmatory research tool.

Where patients develop an apparent HSR and abacavir is stopped, GSK currently recommends that the drug is not re-started even if the patient subsequently tests negative for B*5701. The consequences of re-challenge are so serious that the small chance of a false-negative test cannot be neglected and, although this study did not see any, the chance of rare immunological reactions to abacavir that are not related to B*5701 cannot be ruled out.

Dr Simon Malal, who presented the PREDICT study, concluded that screening patients for B*5701 significantly reduced the risk of patients developing anything resembling a hypersensitivity reaction and eliminated reactions confirmed with a patch test.

He recommended that all abacavir-naïve patients are screened for B*5701 where the facilities to do so exist, but added that “where screening is not readily available, it is reasonable to initiate abacavir with appropriate clinical vigilance. “Clinical vigilance,” he emphasised, “remains the cornerstone of successful abacavir management.”

During the question and answer session that followed, Dr Malal addressed the issue of cost-effectiveness of B*5701 screening in non-white patients. “My belief is that when you have a mixed population – with whites and Latinos and US blacks – it is worth screening everyone because it’s impossible to discriminate against doing screening in someone on the basis of self-identified, or clinician-identified, ethnicity,” he said.

He concluded by noting that, “screening in open studies is eliminating over-diagnosis [of suspected HSR]. This means that all the patients going on the drug can go onto it with confidence, not facing the anxiety that they may have [HSR], and this means that the over-diagnosis rate goes down. If the cost-effectiveness analyses are done on that basis, even in low prevalence groups, it still has value.”