
Full text loading...
Category: Viruses and Viral Pathogenesis
Neurologic Outcomes in HIV Infection: Clinical Events and Surrogate Markers, Page 1 of 2
< Previous page | Next page > /docserver/preview/fulltext/10.1128/9781555815691/9781555813697_Chap08-1.gif /docserver/preview/fulltext/10.1128/9781555815691/9781555813697_Chap08-2.gifAbstract:
This chapter provides a review of the methodology and selection criteria that might be employed in order to identify and validate candidate measures as surrogates of clinical outcome. The analytical procedures outlined here are applicable to a wide range of HIV neurological research but are broad enough to be helpful in general medical investigation. In HIV infection, monitoring of CD4+ T-lymphocyte counts and plasma HIV RNA levels (viral load) is routinely used, rather than observing the occurrence of new opportunistic infections or death. The ability to include uncontrolled studies in meta-analyses evaluating surrogate markers is extremely significant, especially in neuro-AIDS, because the majority of studies that have assessed neurological surrogate markers in HIV have included only a single treatment regimen. The authors simulated a validation study of the screening test in which it was assumed that the true sensitivity and specificity of the screening instrument were at the midpoints of the reported intervals (i.e., 93% and 77%). An overall prevalence of 30% for HIV-associated peripheral neuropathy in the patient population was also assumed. Increasing availability of treatment options in HIV infection has improved the prognosis of many patients infected with HIV. The chapter presents the essential elements for identification and validation of a candidate measure as a surrogate marker. These are distilled into two criteria related to the prognostic ability of the marker for the event of interest and its ability to adequately summarize the treatment effect on the clinical event.
Full text loading...
Valid surrogate marker (solid arrows), in which the treatment effect is manifested through the marker, and incomplete marker (dashed line), in which treatment affects a clinical event through a mechanism that does not involve the marker.
Two alternative meta-analysis situations. (A) Meta-analysis supporting marker surrogacy. (B) Meta-analysis that does not support marker surrogacy, both with regard to the variability among the studies and with respect to the regression line not going through the origin.
Numbers of subjects classified by two tests. T 1 is the gold standard, and T 2 is the diagnostic test under assessment in a conventional validation study (top) and the case-cohort design (bottom). Notice that all subjects who test positive by the screening test (T 2) are tested by the gold standard test (T 1), along with a small number (M) of randomly selected subjects (the subcohort). Thus, it is not known which of the subjects in the overall cohort who tested negative by the gold standard were misclassified by the new test (false positives). Only known group sizes are included in the tables. The numbers of correctly classified subjects are shown in boldface.
PTE explained by three alternative markers
Comparison of case-cohort and conventional estimates of sensitivity and specificity