Lack of Clinical Utility of Cytomegalovirus Blood and Urine Cultures in Patients with HIV Infection

Methods

A list of HIV-infected patients with their initial CMV blood and urine culture results was compiled from the records of the CMV reference laboratory used by the intramural AIDS program at the National Institutes of Health from 1982 to 1989. Included in this list was a group of patients specifically recruited for enrollment in a protocol for the treatment of CMV retinitis. This study has previously been reported [13]. This list was cross-referenced with another patient database file containing CD4 counts done at the reference immunology laboratory. The CD4 count determination done nearest to the date of the CMV culture result was matched with each patient.

Once the list was assembled, individual patient medical records were reviewed and data on the following variables were compiled: age; sex; HIV risk factor; history of interferon, acyclovir, ganciclovir, or foscarnet use before CMV culture; history of recent outpatient fevers or two inpatient temperatures higher than 38 °C within 7 days of CMV culture; history of weight loss of at least 5 kg during the preceding 6 months; date of ophthalmologic or pathologic diagnosis of CMV end-organ disease (either before death or at autopsy); and number of months of follow-up since the time of CMV culture. Diagnoses of CMV disease were defined as follows: For retinitis, the diagnosis was based on ophthalmologic examination or findings at autopsy; for all other end-organ CMV disease, diagnosis was based on the presence of one or more inclusion bodies as shown by histopathologic examination. For patients who were found to have CMV end-organ disease, CMV culture results and CD4 count at the time nearest the diagnosis of end-organ disease were recorded and used in a separate analysis.

Cytomegalovirus isolation was done in duplicate in MRC-5 cells, a human embryonic lung cell line, as previously described [1]. Cultures were done by one microbiologist. All specimens were delivered to the laboratory by special messenger and were usually processed within 1 hour of delivery. All cultures were obtained before the institution of anti-CMV therapy.

Blood Culture

Ten mL of heparinized blood was allowed to settle in an upright position for 30 to 60 minutes at room temperature. The leukocyte-rich plasma was collected and centrifuged at 200 G for 10 minutes. The plasma was removed, and the cell pellet was then resuspended in phosphate-buffered saline and the cells were inoculated onto MRC-5 cell monolayers.

Urine Culture

Five-mL samples were mixed with 5000 units of penicillin, 5000 µg of streptomycin, and 125 mg of fungizone before inoculation onto the monolayers.

All specimens were allowed to absorb the mixture for at least 1 hour at 36 °C, and urine cultures were washed once with phosphate-buffered saline. The cells were maintained in Eagle's minimum essential medium with Earle's salts, 2% fetal calf serum, 100 units of penicillin, 100 mg of streptomycin, and 2.5 mg of fungizone. Media were changed 24 hours after inoculation and twice weekly thereafter for 5 weeks. Blood and urine specimens were collected at various intervals depending on the protocol.

Enumeration of lymphocyte subsets was done on peripheral blood mononuclear cells using fluorescent-activated cell-sorting analysis, after staining with various monoclonal antibodies as previously described [14]. The CD4 lymphocytes were identified phenotypically with OKT4 antibody (Ortho Pharmaceuticals, Raritan, New Jersey).

Statistical Analysis

Relations among CMV culture results, CD4 counts, and end-organ disease were assessed using the chi-square and contingency analyses. The continuity correction was used for 2 × 2 contingency tables. Means are expressed ± SE.

Results

A CD4 count had been done in 393 HIV-infected patients with at least one CMV blood culture result (most also had a urine culture result from the same day). For 28 patients, medical records were missing or contained incomplete data. Of the remaining 365 patients, 322 patients had a CD4 count determination done within 7 days of the CMV culture. These 322 patients were included in the analysis.

The group had a mean age of 36.0 years at the time of CMV culture and included 313 men and 9 women. The HIV risk factor status was assessed as follows: 286 patients (88%) were either homosexual or bisexual, 8 were infected via contaminated blood products, 6 were sexual partners of persons at risk for HIV infection, 4 were intravenous drug users, and 3 were hemophiliacs; the risk factor was uncertain in 15 patients. The mean follow-up for the entire group was 11.8 months (range, 0 to 73 months).

The relation between both CMV blood and urine culture positivity and the subsequent diagnosis of CMV end-organ disease for patients with CD4 counts ≤ 200 cells/mm³ is shown in Figure 1. For this analysis, we used an inception cohort consisting of patients free of CMV disease at the time of the initial CMV culture. Patients with CMV end-organ disease who were recruited for study on the basis of their CMV disease (retinitis [n = 18]) [13] or whose first CMV culture was obtained after the diagnosis of end-organ disease (n = 6) were excluded. Nine of 26 patients (34.6%) with CMV viremia developed end-organ disease compared with 11 of 74 (14.9%) patients without viremia (difference, 19.7% [95% CI, −0.3% to 39.7%]; P = 0.06). Similarly, 15 of 47 patients (31.9%) with CMV viruria developed end-organ disease compared with 4 of 43 patients (9.3%) without viruria (difference, 22.6% [95% CI, 6.7% to 38.5%]; P = 0.02). For CMV viruria, when end-organ disease was diagnosed within 6 months, the difference between the rates of development of disease was 23.1% (95% CI, 8.8% to 37.3%; P = 0.008) when disease was diagnosed before death, the difference was 16.6% (95% CI, 3.4% to 29.9%; P = 0.04) and when disease was diagnosed within 6 months and before death, the difference was 14.7% (95% CI, 3.0% to 26.3%; P = 0.05). For CMV viremia, the difference in rates when end-organ disease was diagnosed within 6 months was 26.5% (95% CI, 8.1% to 44.9%; P = 0.003); when disease was diagnosed both before death and within 6 months, the difference was 19.0% (95% CI, 2.2% to 35.8%; P = 0.01); when end-organ disease was diagnosed before death alone, the difference was 13.6% (95% CI, −3.8% to 31.1%; P = 0.15).

View larger version:

• In this window
• In a new window

Figure 1. The percent of patients with end-organ disease who had a positive blood culture ( = 26), a negative blood culture ( = 74), or a positive urine culture ( = 47), or a negative urine culture ( = 43) for CMV is shown. Cytomegalovirus (CMV) end-organ disease according to CMV blood and urine culture results in human immunodeficiency virus-infected patients with CD4 counts ≤ 200/mm³ and without baseline CMV end-organ disease.nnnn

Cytomegalovirus viremia did not correlate with the presence of either fever or weight loss when patients with CD4 lymphocyte counts ≤ 200/mm³ were considered. Ten of 41 patients (24.4%) with CMV viremia had fevers compared with 21 of 83 patients (25.3%) without viremia (P > 0.2). Similarly, 15 of 41 patients (36.5%) with CMV viremia experienced significant weight loss compared with 25 of 83 patients (30.1%) without viremia (P > 0.2). No relations could be determined for antiviral therapy because few patients were receiving acyclovir and no patients were receiving ganciclovir or foscarnet at the time of CMV blood or urine culture.

Overall, 51 patients were diagnosed with CMV end-organ disease. Of these patients, 19 with retinitis have been previously described [13]. Data from autopsy studies in the current series have also been reported [11]. Retinitis was diagnosed in 28 patients and was the most common end-organ disease, followed by pneumonitis (n = 14) and colitis (n = 7). Isolated adrenal disease was found in 2 patients at autopsy, and 13 patients overall had multiorgan involvement. Among the subgroup of patients without CMV end-organ disease at baseline (n = 25), pneumonitis later developed in 12, retinitis in 7, and colitis in 4. Two patients had adrenal disease at autopsy, and 11 patients overall had multiorgan involvement at autopsy. Specific end-organ involvement did not differ between patients with CMV viremia and those without viremia.

In 29 patients who had documented CMV end-organ disease and a CMV culture that was obtained within 7 days of the diagnosis, the results of blood and urine cultures were not particularly sensitive or specific (Table 1). Overall, using the single initial culture as the standard, we found that only 55% of patients with documented CMV disease were viremic and that 88% were viruric. Nineteen of these patients composed a subset of a group of 24 patients with retinitis who took part in a careful, prospective trial assessing the efficacy of phosphonoformate [13]. Only 15 of those 24 patients (61%) were viremic. All 24 patients, however, were viruric. Thus, the presence of viremia was neither a sensitive nor a specific diagnostic marker. Viruria was sensitive in identifying a population that could have CMV disease, but it was a highly nonspecific test.

Forty-four of 322 patients (13.7%) had CMV viremia, and 87 of 293 patients (29.7%) who had CMV urine and blood cultures done on the same day had viruria. Mean age, sex distribution, and HIV risk factor did not differ when patients were grouped according to CMV blood or urine culture status.

As shown in Figure 2, CMV blood and urine culture results were stratified according to CD4 count for the 293 patients whose CMV urine and blood cultures were done on the same day. Proportions of persons with blood and urine cultures positive for CMV varied inversely with CD4 count, (P = 0.0001). Of the 35 patients who had positive CMV blood cultures and for whom urine culture results from the same day were available, 30 (85.7%) also had positive urine cultures; of the 258 patients who had negative blood cultures, 201 (77.9%) also had negative urine cultures. Cytomegalovirus viremia was strongly associated with CMV viruria (P = 0.0001). Similarly, CMV urine culture negativity was highly associated with CMV blood culture negativity. However, CMV viruria was poorly predictive of CMV viremia. Of the 87 patients who were urine culture positive, only 30 (34.5%) were blood culture positive. Of the 64 patients who had CMV viruria and CD4 counts ≤ 200 cells/mm³, only 28 (43.8%) were blood culture positive.

View larger version:

• In this window
• In a new window

Figure 2. Results are shown for 293 patients with cytomegalovirus blood and urine cultures done on the same day. Cytomegalovirus blood and urine culture results according to CD4 count.

The time to development of CMV end-organ disease among the 25 patients free of CMV disease at the time of the initial culture determination is shown in Figure 3; patients are grouped according to CMV blood and urine culture status and CD4 count (≤ or > 100/mm³). Patients who had CMV viremia or viruria developed end-organ disease sooner (84.1 ± 38.6 days and 95.9 ± 30.5 days; respectively) than those who did not have viremia (260.3 ± 70.3 days) or viruria (361.9 ± 130.4 days). However, independent of the CMV culture result, when this same group of 25 patients was grouped according to CD4 count, patients with CD4 counts ≤ 100/mm³ developed end-organ disease sooner (85 ± 18.4 days) than those with counts above 100/mm³ (335.6 ± 110.4 days).

View larger version:

• In this window
• In a new window

Figure 3. Individuals are grouped according to CMV culture result. A separate analysis groups individuals according to CD4 count greater and less than 100/mm , independent of CMV culture result. Ten patients were blood-culture–positive; 15, blood-culture–negative; 17, urine-culture–positive; and 7, urine-culture–negative. Sixteen patients had CD4 cell counts ≤ 100/mm , and 9 patients had counts greater than 100/mm . Boxes represent mean ± SE. Time to development of cytomegalovirus (CMV) end-organ disease according to CMV blood and urine culture results among 25 patients free of CMV disease at the time of initial culture.333

Discussion

In our study of 322 patients with HIV infection, both CMV viremia and CMV viruria were found to have some prognostic utility. Among patients with circulating CD4 lymphocyte counts ≤ 200/mm³, viremic patients or viruric patients were statistically more likely to develop end-organ disease than those who were culture negative. However, the positive predictive values of CMV viremia (35%) and viruria (28%) for the development of end-organ disease within 6 months were relatively low. Consequently, the use of blood and urine culture results to predict the development of CMV end-organ disease adds relatively little diagnostic information.

In clinical practice, health care providers often obtain CMV blood cultures for diagnostic purposes in HIV-infected patients with unexplained constitutional symptoms. In our study, neither fever nor substantial weight loss was more frequent in patients with viremia than in those without viremia. Thus, obtaining CMV blood cultures as part of an evaluation may have little diagnostic utility. As part of the workup of an HIV-infected patient with end-organ disease such as retinitis, pneumonitis, or colitis, health care workers also order CMV blood cultures to help establish a diagnosis. Even in this population, a negative blood culture does not rule out CMV as the cause, and a positive blood or urine culture does not provide a specific diagnosis. Thus, a positive blood or urine culture for CMV is likely to be more a reflection of the patient's immunologic status than a clue to the cause of the disease. Cytomegalovirus retinitis can usually be diagnosed by ophthalmologic examination. For patients with esophagitis, colitis, or pneumonitis, the histologic evaluation of a biopsy specimen to demonstrate characteristic intracytoplasmic and intranuclear inclusions remains the best tool for determining whether CMV is causing the organ dysfunction, although considerable uncertainties remain in assessing how representative a given biopsy sample is. Quantitative CMV cultures or quantitative assays for circulating antigen may ultimately be useful, but these techniques have not yet been validated.

In our study, a clear association was found between CMV culture results and the circulating CD4 count. Of 43 viremic patients, 38 had CD4 lymphocyte counts below 100/mm³ and, indeed, many had counts below 50/mm³. Cytomegalovirus blood cultures were positive in only one patient with a CD4 count of more than 300/mm³. Cytomegalovirus was found in the urine of patients with a broader range of CD4 lymphocyte counts, which is not surprising because healthy, immunologically sound persons without HIV infection can shed CMV in urine [15, 16]. Nevertheless, CMV viruria occurred more frequently in patients with low CD4 lymphocyte counts. Seventy of 87 viruric patients (80%) had CD4 lymphocyte counts below 300/mm³, and 53 of 87 patients (60%) had CD4 lymphocyte counts below 100/mm³.

Some significant associations were seen when CMV blood culture results were compared with simultaneous urine culture results. When the CMV blood culture was positive, the urine culture was positive in most cases. Similarly, when the blood culture was negative, the urine culture was usually negative. In contrast, whereas most patients who were urine culture negative were also blood culture negative (201 of 206 patients), only 35% of patients who were urine culture positive were also blood culture positive. Even when the subgroup of patients with CD4 counts ≤ 200/mm³ was analyzed, only 44% of patients who were urine culture positive were also blood culture positive.

Our study assessed only the initial blood and urine cultures or the culture nearest in date to the documentation of end-organ disease. Serial, prospective sampling would probably have identified additional viremic or viruric patients, as viral shedding may not be continuous. Because specimens were not obtained from every patient at regular intervals, it seemed reasonable to focus on the initial specimen. It may be argued that specimen handling and laboratory processing might have varied, placing in question the accuracy of a single culture determination from each patient. However, one microbiologist did all the cultures in the 8-year study period on specially delivered specimens, and therefore the yield of cultures in this setting was probably better than in most routine settings.

In conclusion, our data suggest that for diagnostic purposes, CMV is not invariably present in the blood of patients with clinically apparent CMV disease. Further, although most patients with CMV disease are viruric, viruria is highly nonspecific. At best, a negative urine culture may be diagnostically suggestive in excluding CMV disease but is clearly not definitive for ruling out the presence of clinically active CMV disease. Cytomegalovirus culture results do not correlate well with the presence or absence of fever or weight loss and are poorly predictive of end-organ disease; rather, they appear to be markers for the relative degree of immunosuppression because CMV viremia and viruria occur much more commonly when HIV-infected patients have low circulating CD4 lymphocyte counts. Thus, obtaining CMV blood or urine cultures as part of the evaluation of HIV-infected patients has little prognostic or diagnostic utility. Also, given the high cost of such cultures ($53.60 for a single blood culture, and $69.50 for a single urine culture at a local commercial laboratory), they should be used primarily in research or in special cases that warrant drug susceptibility testing.

The contents of this article do not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade name, commercial products, or organizations imply endorsement by the U.S. Government.