The incidence of illness and death at relatively high CD4 cell counts leapt to prominence as an issue at the Thirteenth Conference on Retroviruses and Opportunistic Infections in Denver last week with the announcement of the findings from the prematurely terminated SMART study and other trials of structured treatment interruptions, as reported here.
Although most of the excess mortality in the SMART study was due to non-AIDS-defining conditions, certain opportunistic infections such as oesophagal candidiasis (thrush) were also prominent at higher-than-expected CD4 cell counts.
One poster at the conference attempted to quantify how often six of the most common AIDS-defining opportunistic infections (OIs) occurred in Europeans at CD4 cell counts normally considered high enough to guard against them.
Crucially, it found that although opportunistic infections at high CD4 cell counts are quite rare, they are over twice as common in people not taking HAART as in those taking it, regardless of CD4 count – a finding in line with the SMART study.
The EuroSIDA Cohort is a prospective cohort of patients at 72 treatment centres in Europe, Israel and Argentina. It recruited its first members in May 1994 so represents nearly twelve years of data.
The analysis presented at the conference did not attempt to cover all OIs but picked six and stratified them according to the CD4 count above which they would not normally be expected to occur. They were:
- Group 1: OIs not expected at CD4 cell counts over 100 copies/mm3: cytomegalovirus (CMV), Mycobacterium avium complex (MAC), toxoplasmosis.
- Group 2: not expected at CD4 cell counts over 200 copies/mm3: Pneumocystis jiroveci (formerly carinii) pneumonia (PCP), oesophageal candidiasis.
- Group 3: not expected at CD4 cell counts over 300 copies/mm3: active tuberculosis (TB).
It found 212 cases of such OIs in a cohort now numbering 11,229 patients, with 67 in group one, 112 in group 2, and 22 in group 3.
This does not simply translate into 1.88% of patients getting OIs at high CD4 cell counts. On the one hand some patients may have had more than one OI, which would tend to overstate the per-patient incidence; on the other hand the cohort is continually expanding and there were fewer patients in earlier years, which would tend to understate the incidence.
The researchers were able to calculate the annual per-patient incidence of OIs and also the incidence on and off HAART thus:
Incidence per 1,000 patient-years | Group one | Group two | Group three |
| Overall | 1.7 | 3.8 | 1.3 |
| On HAART | 1.0 | 2.5 | 0.9 |
| Off HAART | 4.0 | 8.0 | 3.0 |
They calculated that in groups 1 and 2 the initiation of HAART cut the likelihood of an OI at higher-than-expected CD4 counts by 54%. A related finding was that people with viral loads over 1,000 were 4.4 times as likely to get an OI in group 1 and 2.63 times as likely to get an OI in group 2, compared with people with viral loads under 1,000.
They also calculated that if the CD4 count fell by 50%, the chances of getting an OI in group 1 increased 5.6 times and fourfold in group 2.
This can be illustrated by the OIs in group 1:
| CD4 Count | Incidence per 1,000 patient-years |
| 100-199 | 8 |
| 200-299 | 2 |
| 300-399 | 1 |
| 400-499 | 0.4 |
| ≥500 | 0.1 |
In line with the SMART study findings, CD4 nadir (lowest-ever count) and baseline CD4 (at entry to the cohort) had no influence at all on the likelihood of developing OIs.
In group 3, the only influence on whether people got TB was whether they were injecting drug users: IDUs were twice as likely to get it.
One finding not highlighted on the poster but evident from the figures was that the excess incidence in group 2 was entirely due to oesophageal candidiasis. There were 84 cases of this compared with 28 of PCP, and it was four times as common as any of the other OIs. It was also the most common OI reported in the SMART study. This may indicate that the CD4 threshold above which it ‘ought’ not to occur needs to be revised.
The researchers comment: “Opportunistic infections may be present in patients with clinically suggestive symptoms if the latest CD4 count is less than twofold above the usual CD4 threshold for that OI, and even more so if the patient has not initiated HAART.
“Tuberculosis at a CD4 count over 300 should particularly be suspected in case of compatible symptoms in a person using intravenous drugs.”
Podlekareva D et al. Factors associated with the development of opportunistic infections in HIV-1-infected adults with high CD4 cell counts in the EuroSIDA study. Thirteenth Conference on Retroviruses and Opportunistic Infections, Denver, abstract 783, 2006.