In this study we examined the utility of plasma suPAR concentrations for predicting all-cause mortality among HIV-infected patients with advanced immunodeficiency early in an ART programme in South Africa. Our observations add to previous findings from Europe that plasma suPAR is an independent predictor of mortality in untreated cohorts of HIV-infected persons [14, 15]. We also found that mortality was more strongly associated with suPAR concentration than with any other risk factor including CD4 cell count and WHO clinical stage. However, despite these strong associations, there was no useful cut-point in the distribution of suPAR concentrations that could discriminate adequately between groups with higher and lower mortality risk. Thus, the marker was unable to serve as a useful adjunctive tool to assess short-term mortality risk in this patient population.
SuPAR is a component of the plasminogen activation system, which comprises urokinase-type plasminogen activator (uPA) and its receptor (uPAR) . uPAR is expressed on a variety of different immune cells and vascular endothelial cells. It is involved in the recruitment of leukocytes from the circulation to extravascular sites of inflammation through regulatory effects on pericellular proteolysis, cell adhesion, chemotaxis, and signal transduction [26, 27]. suPAR is generated by either proteolytic cleavage or shedding from cells. Serum concentrations are increased and prognostic in a variety of inflammatory and neoplastic disease states, including HIV infection, tuberculosis, sepsis, rheumatoid arthritis and certain malignancies and suPAR concentrations reflect prognosis in these disease states [14, 15, 17, 28–30].
In European cohorts of HIV-infected patients plasma suPAR concentrations were higher among those with more advanced clinical stages of disease [14, 15]. This is likely to reflect the fact that systemic immune activation increases with HIV disease progression and is central to the pathogenesis of the infection . In this study we also found that higher suPAR concentrations were significantly associated with advanced WHO stage of disease and lower blood CD4 cell counts. Absolute levels reported in this patient population are consistent with those found in previous studies [14, 15]. High suPAR concentrations were also found to be weakly associated with male sex. This has not previously been reported although it is interesting that male sex was an independent risk factor for early mortality in several African ART cohorts [4, 11, 32]. Although tuberculosis has been found to be associated with increased supAR levels  there was no such association in this study. This may reflect the fact that those who were free of tuberculosis may have had a high frequency of other opportunistic infections, which also increase suPAR concentrations, thereby masking any association with TB.
The association between high suPAR concentrations and mortality was highly statistically significant both at a group level (Figure 1) and in multivariate analysis (Table 3). However, the distributions of suPAR concentrations in those who survived and those who died were very broadly overlapping (Figure 1). This is likely to reflect the fact that suPAR is a non-specific inflammatory marker with a continuous distribution. As a result, using ROC analysis it was not possible to identify a useful cut-point that provided adequate sensitivity and specificity to enable the assay to be used to discriminate those at highest risk of death. This assay could therefore not be used to triage patients according to mortality risk as they enrolled into this ART programme.
A strength of this study is that we studied patients in a community-based service that is likely to be representative of other public sector ART clinics elsewhere in sub-Saharan Africa. Moreover, a cohort rather than case control design reduced potential for patient selection bias. Completeness of outcome data in this cohort was high due to active community-based follow-up of patients using peer counsellors. Some patients who were lost to the programme for reasons other than death may have subsequently died. However, these patients were categorised separately in the group analysis and their median suPAR concentration did not differ from that of the group who remained alive. Previous studies have only examined associations between plasma suPAR concentrations and prognosis in HIV-infected patients at a group level. This is the first study to evaluate the practical clinical utility of this marker in a clinical setting.
The study assessed short-term mortality risk over a median of 5 months from enrolment. An important reason for this is that we have previously reported that deaths in this period are strongly associated with baseline patients characteristics  whereas deaths occurring beyond this period are only associated with response to ART. Previous studies of the prognostic value of suPAR concentration have reported on much longer durations of follow-up of 3 years  and 5 years  in natural history cohorts. Plasma suPAR concentrations may have greater prognostic value in assessing long-term mortality risk when assessed in patients with less advanced immunodeficiency. However, short-term mortality risk is of greatest relevance to clinicians making clinical decisions and for developing treatment guidelines [33, 34].
Only patients with advanced immunodeficiency were included and the restricted cohort composition may have diminished the association between mortality and risk factors such as CD4 cell count. This would also explain the lack of association with plasma viral load . Despite the cohort composition, the association between mortality and plasma suPAR concentrations was nevertheless very strong at a group level. The data from this study are relevant to patients with advanced disease who are already eligible for ART and cannot be generalised to those with earlier disease. Whether this assay can provide useful prognostic information relevant to when ART should be started needs to be assessed in future studies. This will depend on whether the distribution of suPAR concentrations in patients with earlier stages of disease has a discriminatory threshold.
In conclusion, at a group level plasma suPAR concentrations are strongly predictive of mortality among HIV-infected patients in sub-Saharan Africa, confirming the findings of previous studies in Europe. However, the lack of a cut-point that provides adequate sensitivity and specificity to predict mortality in this patient population prevented this assay from providing clinically useful prognostic information for individual patient management.