Abacavir-based regimens are slightly more cost-effective than tenofovir-based ones, a US mathematical modelling study published in AIDS has concluded. But this is only the case if patients are tested at the outset for the abacavir hypersensitivity reaction, a potentially fatal allergy to the drug.
The study found that testing for the HLA-B*5701 genetic mutation, which confers hypersensitivity to abacavir, and using tenofovir-based regimens only with those who have it, was the most cost-effective option when compared with using tenofovir from the start with all patients, or not testing for B*5701 and clinically managing the hypersensitivity reactions that arose. The strategy would also produce a small increase in patient life expectancy.
However the margins of difference are extremely small. In the primary model tested, the lifetime estimated cost for treatment and care of a patient starting with abacavir plus the B*5701 test was $472,320. The cost for the same patient starting on tenofovir was $30 less.
Similarly the estimated Quality-Adjusted Life Expectancy (QALE) of a patient started on abacavir plus B*5701 testing was 194.79 months or 16.23 years (this is the estimated amount of illness-free years, and is not the same as the actual life expectancy, which was 30.93 years). The average QALE of a patient starting on tenofovir would be 194.71 months – a difference of 2.4 days.
The average annual cost of the abacavir regimen (abacavir, 3TC, efavirenz), based upon actual prices paid in the USA, was estimated as $13,620 (£7,439 at today’s rate) and that of the alternative, tenofovir, FTC, efavirenz as $13,668. If this $48 dollar difference were eliminated, starting on tenofovir would be more cost-effective.
The cost per Quality-Adjusted Life Year (QALY) saved of the abacavir-plus-B*5701 regimen is $36,700 (about £20,000 at today’s exchange rate). This brings it comfortably within the $50,000 to $100,000 maximum boundary for what US health authorities regard as cost-effective. All other regimens tested were less cost-effective.
The reason why tenofovir is slightly less effective is because the model assumes that a certain number of patients will develop tenofovir-related kidney toxicity and that if they do they will switch to AZT-based regimens, which have their own toxicities. Switching for tenofovir toxicity to abacavir (with or without B*5701 testing) would be the most expensive option of all, costing $230 more than starting with abacavir-plus-test.
The model does not incorporate new data suggesting that abacavir may increase the risk of heart attacks (see this report). It also assumes that for patients overall, abacavir (plus 3TC) is as potent as tenofovir (plus FTC). If abacavir is in reality even 1% less effective than tenofovir (in terms of the number of patients achieving viral loads under 400), then tenofovir-based regimens become more cost-effective.
It does incorporate the possibility that abacavir may work less well in patients starting with viral loads over 100,000, as another recent study found (see this report) and finds that restricting abacavir to patients with baseline viral loads under 100,000 would save some money, with a cost per QALY of $35,400.
The model started with an idealised average patient (age 36, 74% likely to be male, baseline CD4 count 276). It took the proportion of patients who were positive for the HLA-B*5701 gene as 5.7%, the figure seen in PREDICT-1, a large international trial of B*5701 testing (see this report). Varying this for the ethnic mix seen in US patients produced a B*5701 prevalence of 4.1% and an increased cost per QALY for the regimen of $45,200.
The model uses US patient data that shows roughly 60% of hypersensitivity reactions (HSRs) are mild, 40% severe enough to need hospitalisation, and that 0.7% are fatal (one in 143 HSRs or one death due to HSR in every 5232 untested patients).
The model also uses data from PREDICT-1 showing that hypersensitivity reactions occur in 48% of patients who are B*5701 positive (2.3% of all patients) and that B*5701 testing prevents all true HSRs. The model also assumes that the prevalence of unconfirmed HSRs, based on suggestive symptoms alone, in patients on abacavir would be 5.1% and that using the B*5701 test would reduce this to 3.4%. This is probably a conservative estimate, because the PREDICT-1 study was blinded, so physicians did not know if patients had taken the B*5701 test. If doctors assumed that no patient who was B*5701 negative could have a “true” hypersensitivity reaction, thereby eliminating all immunologically unconfirmed diagnoses, this would bring the cost per QALY of the abacavir-plus B*5701 regimen down by over $3000 to $33,500.
Schackman BR et al. The cost-effectiveness of HLA-B*5701 genetic screening to guide initial antiretroviral therapy for HIV AIDS 22(15):2025-2033. 2008