Biomarkers associated with mortality: long-term follow up from SMART

Nathan Geffen, i-Base and TAC

A poster at IAS2009 by Nick Paton and the INSIGHT SMART Study Group presented long-term follow-up data from the SMART study on biomarkers associated with mortality. [1] This analysis extended an earlier nested case-controlled study of the association between biomarkers and mortality.

The earlier study identified all 85 patients who had died up to 11 January 2006, ie the date that enrollment into SMART was stopped. Each death was matched to two controls by country, age, gender and randomisation date. The study evaluated four inflammatory markers, hsCRP (C-reactive protein measured using the highly sensitive test), interleukin-6 (IL-6), Serum amyloid A and serum amyloid P. It also examined three coagulation markers, D-dimer, PA1-1 and Prothombin fragment 1+2 (F1.2). Three markers, hs-CRP, IL-6 and D-dimer, were found to have a statistically significant association with mortality on both adjusted and unadjusted odd ratios.

Table 1: Case-controlled odd ratios by baseline biomarker levels

Marker Unadjusted OR (4th /1st quartile) P-value Adjusted OR (4th /1st quartile) P-value
CRP 2.0 0.05 2.8 0.03
IL-6 8.3 <0.0001 11.8 <0.0001
D-dimer 12.4 <0.0001 26.5 <0.0001

The extended analysis reported at IAS2009 included all deaths up to 11 July 2007 in order to determine whether the association between these biomarkers and mortality persists. There were 167 deaths in the SMART cohort up to that point, 85 before the protocol modification (to offer all patients continuous treatment) and 82 post-modification. For this analysis, the deaths were however divided into early (<=2 years after randomisation, n = 95) or late (>2 years, n = 71). Two cases were matched to each death as in the baseline study. The baseline values of two of the three biomarkers (IL-6 and D-dimer) continued to be statistically significant predictors of late deaths and there was a trend for CRP to be a predictor of late deaths (see Table 2).

Table 2: Baseline biomarker levels and risk of death

Marker Early (2yrs) Deaths Controls Adjusted OR* p-value
No. Median No. Median
Hs-CRP (µg/ml) Early 96 3.13 188 2.08 2.8 0.009
Late 71 3.09 137 1.93 2.8 0.08
IL-6 (pg/ml) Early 92 3.58 184 2.14 5.9 <0.0001
Late 67 3.72 133 2.33 6.4 0.004
D-dimer (µg/ml) Early 94 0.45 188 0.24 7.3 <0.0001
Late 69 0.31 138 0.24 8.3 0.002

* 4th/1st quartile

Greater predictors of risk than other factors

Table 3 shows some of the other risk factors for deaths that have been found in SMART (note that where p-values show non-significance, the factor can still be significant when the early and late groups are counted together).

Table 3: Risk factors associated with mortality in SMART

Risk factor Deaths (%) Controls (%) p
Hepatitis B or C Early 38.9 20.7 0.002
Late 46.5 19.3 0.0001
Current smoker Early 50.5 34.0 0.006
Late 64.8 38.6 0.005
Diabetes Early 18.9 10.6 0.07
Late 22.5 13.6 0.08
Blood pressure drugs Early 37.9 25.0 0.02
Late 38.0 23.6 0.02
Prior CVD history Early 10.5 5.9 0.01
Late 15.5 2.1 0.002
Total/HDL cholesterol Early 4.4 4.7 0.06
Late 4.8 4.8 0.99
Treatment group (% on structured breaks) Early 63.2 52.1 0.09
Late 59.1 50.7 0.27

The authors noted that IL-6, hs-CRP and D-dimer are associated with greater risk of mortality than smoking and diabetes and about an equivalent risk of prior cardiovascular disease. They conclude that interventions to decrease inflammatory and coagulation pathway activation may be of long-term benefit for people with HIV.


The association between mortality and hs-CRP, IL-6 and D-dimer is significant, even after long-term follow-up and the termination of the structured treatment interruption arm. This highlights the importance of further research on whether anti-inflammatory medicines will have an additional role in HIV management of high-risk patients.

It would be interesting to know the association between these biomarkers and mortality is in the uninfected population. If they are similarly prognostic, then the question is how much of the associations seen in SMART are HIV-specific.

The role of early HAART in mitigating the association with mortality also needs to be determined.


  1. Paton N. Association between activation of inflammatory and coagulation pathways and mortality during long-term follow up in SMART. 5th IAS Conference on HIV Pathogenesis, Treatment and Prevention. 19-22 July 2009, Cape Town. Oral abstract MOPEA034.

Links to other websites are current at date of posting but not maintained.