HIV-1 messenger RNA in peripheral blood mononuclear cells as an early marker of risk for progression to AIDS

Authors Saksela K, Stevens C, Rubinstein P, Taylor P, Baltimore D.
Source Annals of Internal Medicine. 123:641-8. November 1, 1995.
Institutions Rockefeller University, New York Blood Center, M.I.T.
Support NIH; Aaron Diamond Foundation for AIDS research.



This study was designed to look at HIV mRNA levels in peripheral blood mononuclear cells as a marker for disease progression. The patient population from which this study derived was the Prospective AIDS study which enrolled patients starting in 1984 with follow-up every 4.5 months for up to 8 years. Blood samples were drawn for various immunologic markers and peripheral mononuclear cells were cryopreserved. For the present study, 150 cryopreserved specimens were selected at random from samples obtained in 1984 or early 1985 from 327 men who had not yet developped an AIDS defining illness.


Both unspliced and multiply spliced HIV mRNA levels were determined from the cryopreserved samples by PCR; these were subsequently found to be highly correlated with each other, and only the multiply spliced mRNA levels were used in the analysis. Measurements of other markers of disease progression (such as CD4+ counts) were taken as the average of the measurement at the time of the mRNA sample and the measurement at a preceding and following sample (Example: a patient seen in January, June and October -- mRNA measured from cryopreserved specimen collected in June, CD4+ count taken as average of January, June and October measurements).

Subjects were classified into four quartiles of CD4+ counts (68-443, 443-624, 625-820 and 821-1923; 37 or 38 subjects in each quartile). They were also classified into 5 groups based on the mRNA levels: one group of 17 patients with undetectable mRNA levels and the remaining 133 divided into 4 equal-sized groups based on the quantity of mRNA.


CD4+ cell counts were inversely correlated with mRNA levels, as was expected. This correlation was not absolute, however, and there was much overlap between the groups (with some in the highest CD4+ category having mRNA levels typical for the lowest CD4+ category). mRNA levels turned out to be highly predictive of subsequent development of AIDS. After 3 years, AIDS had developed in 69% of men in the highest mRNA category, 34% in the next category, 6% in the two lowest expressor categories and none in the non-expressor category. After 7 years, only 9% in the highest two categories were alive and free of AIDS, compared with 44% in the two lowest expressor category. In the non-expressor category, only one patient developed AIDS, and retesting of the samples obtained from that individual before and after the index sample yielded measurable mRNA, suggesting that the negative value was artefactual.

Although CD4+ cell counts also correlated with subsequent progression to AIDS, they did not help identify any subgroup of long-term non-progressors. Furthermore, in multivariate analysis, mRNA levels added significant prognostic information to the CD4+ counts.


One of the most interesting aspects of this study, as the authors point out, is the ability of peripheral mononuclear cell HIV mRNA to predict a subgroup of patients who are long-term non-progressors. Determining the importance of host-factors and viral factors in this non-progression will be important. Plasma HIV RNA load after seroconversion was recently found to correlate with clinical outcome, and plasma HIV RNA is easier to measure, but plasma HIV RNA does not appear as predictive when measured later in the course of the disease. The exact clinical role of peripheral mononuclear cell HIV mRNA determination remains to be seen but appears promising.

November 27, 1995


References related to this article from the NLM's PubMed database.

Reader Comments

Subject: HIV mRNA and progression to AIDS
Date: Wed, 29 Nov 1995 21:44:17 -0500

Although markers for disease progression are great for research purposes, I feel they are used much too often in clinical practice. The value in checking CD4 counts has not been in knowing how much time one had left, but in knowing when to start various interventions. Similarly, with the mRNA assays, we should be doing studies to find out if various interventions (drugs, etc.) used to decrease viral load have a measureable impact on clinical parameters such as survival and AIDS-free survival. The early identification of those with low viral loads may be of use to researchers trying to figure out how their immune system accomplishes this.

Roger Spitzer MD

I agree with Dr. Spitzer...I have measured HIV-RNA "viral loads" in several patients (often at their request) but do not know really what to do with that info. In a couple of patients who were clinically doing well but had low CD-4 counts, I was somewhat reassured that the RNA titer was not all that high, either...however, whereas I have a lot of experience with CD-4 counts, I really don't have a feel yet for the HIV-RNA...what is really high, what is worrisome, what is reassuring that things are OK for the present. I have been sending more of these to try to get a feel of what they mean clinically...I will certainly be interested in more articles about "viral load" measurements.

    Received 11/29/95 (the way the form was set up, I didn't get a return address) --mj

Date: Sat, 17 Feb 1996
Subject: Viral load Studies

I find viral load testing very helpful. The patient who is reluctant to take a drug can be convinced if the load is high, or for the patient with a the stable CD4 count with low viral load, you intervene when the load increases, which tends to predate the drop in CD4 count. The major obstacle is cost and reimbursement. Hopefully the recent reports from the Washington retrovirus Conference will improve that problem.

Ronald Hirsch, MD, FACP
Key Medical Group
Lake In The Hills, IL

Date: Sun, 03 Mar 1996
From: David Winsemius MD MPH <>
Subject: HIV RNA surrogate markers

Don't we remember all the efforts on various cholesterol-lowering agents, many of which have actually increased mortality (eg. d-thyroxine, clofibrate)? The paradigm of (X is associated with death, so lower X) fails for many human diseases. Why can't we put aside these surrogate endpoints, especially for clinical, day-to-day use? The researchers may get some insights from this kind of study, but the human payoff is in the years-of-life saved in a real clinical trial, not in some petri dish or EMIT assay.

    The problem with HIV is that new therapies and combination therapies are appearing and are being used all the time, and it can take years to detect effects on mortality. Some sort of surrogate validation is needed to try to weed out the ineffective treatments, even if these markers are imperfect. Your point is well-taken, however, and we need to keep reminding ourselves that surrogate markers are not what count, in final analysis. -- mj

August 28, 1997

It has been almost two years since this study was published. In doing a quick literature search, it would appear that measuring HIV mRNA in peripheral blood mononuclear cells has not become a common test. In fact, I couldn't find any recent articles on this topic.

On the other hand, the use of plasma viral load measurements for gauging disease prognosis and response to therapy has become quite accepted. Presumably, these measurements (plasma viral load by PCR) are simpler and cheaper than the technique discussed in this article while yielding similar information. -- mj

Submit a comment

Journal Club homepage

Site Meter