The study was reviewed and approved by the Cleveland Clinic Institutional Review Board. Informed consent was obtained from a parent or guardian of each person <18 years of age, or directly from persons >18 years of age; assent was obtained from children 7–17 years of age. Patients were enrolled from August 2004 through April 2005. Enrolled children were receiving treatment for a malignancy or were receiving immunosuppressive therapy after solid-organ transplantation (SOT). Inclusion criteria were an age of newborn to 21 years and new onset of fever with an oral or rectal temperature >38°C or an axillary temperature >37.5°C. At enrollment, we collected a blood specimen and information on age, sex, underlying illness and diagnosis, acute symptoms accompanying fever, and details about immune suppression (chemotherapy regimen, radiation therapy, immune modulators, and steroids). Gastrointestinal symptoms were defined as either vomiting or diarrhea. Clinical laboratory data collected included complete blood counts, renal and liver function test results, bacterial blood cultures, as well as results of testing conducted for evaluation of fever, such as respiratory virus analysis, stool studies, chest radiographs, and computed tomography scans.

Whole blood was collected in EDTA tubes and stored at 4°C until processed. A leukocyte pellet was collected from 1 mL of whole blood; 400 μL of plasma was preserved, 200 μL for PCR and 200 μL for serologic analysis. Pelleted leukocytes were added to MagnaPure LC lysis buffer (Roche Diagnostics, Indianapolis, IN, USA) and stored at −20°C.

CMV culture was conducted with pelleted leukocytes by using conventional MRC-5 cell shell vial cultures (Diagnostic Hybrids, Athens, OH, USA). Immune fluorescence testing was conducted after 48 hours by using a monoclonal antibody directed against the CMV immediate early protein (Chemicon, Temecula, CA, USA). Shell vial cultures were obtained for the first 11 patients enrolled; all cultures were negative. Because shell vial culture uses a large number of leukocytes and we were studying a neutropenic population, this procedure was discontinued to preserve cells for more sensitive PCR analysis (26). Purified lymphocytes were cultured to detect HHV-6 or HHV-7 infections as described (27,28). Cytopathic effect was recognized as enlarged and refractile cells, with confirmatory testing by PCR.

Template DNA was extracted by using the MagnaPure automated extraction method (Roche Diagnostics) and eluted into 50 μL of buffer. A total of 5 μL of eluted DNA was used in each PCR. The PCRs contained DNA extracted from the equivalent of ≈20 μL of plasma, leukocytes present in 100 μL of whole blood, and 20 μL of whole blood. For detection of CMV and HHV-6 variants A and B, the Artus CMV RG PCR ASR/Real Art/Qiagen RealArt CMV LC Assay and the RealArt/Qiagen HHV-6 A/B LC Assay (QIAGEN, Hamburg, Germany) were used. A co-amplified internal control was used to identify PCR inhibition. To differentiate between HHV-6 variants, the system uses the specific melting temperatures of the probes. Detection of HHV-7 was conducted by using previously described primers and hybridization probes (29). Real-time PCR was performed on a LightCycler system (Roche Diagnostics).

CMV serologic analysis was performed by using the VIDAS CMV IgG Assay (bioMérieux, Durham, NC, USA) in an automated instrument, according to the manufacturer’s protocol. Analyses for HHV-6 (using HHV-6A–infected cells) and HHV-7 were conducted by using conventional indirect immunofluorescence with commercial slides (Panbio Ltd, Columbia, MD, USA). Avidity of antibody to HHV-6 was tested as described by Ward et al. (30).

The exact unconditional test was used to compare groups on viral activity (31). All tests were 2-tailed and performed at a significance level of 0.05. Analyses were performed by using StatXact PROCs version 6.3 (Cytel Software Corp., Cambridge, MA, USA) and SAS version 9.1 (SAS Institute Inc., Cary, NC, USA). Clinical associations with β-herpesvirus infection were based on symptoms and concomitant laboratory findings in patients who had no other identifiable cause for their febrile illness.

Forty-one children were enrolled, from whom 39 evaluable specimens were available (Table 1). Thirty of the 39 children had a malignancy (our primary interest) and 9 were organ transplant recipients who were included for comparison. All cancer patients were receiving active therapy, except for 1 child who was included in the study because she had been receiving chemotherapy for 2.5 years, which ended 3 months before study enrollment. Several patients had symptoms in addition to fever, mostly upper respiratory or gastrointestinal. A source of fever was identified for 5 (17%) of 30 cancer patients and 2 (22%) of 9 organ transplant recipients. Two cancer patients had bacteremia (Escherichia coli, coagulase-negative staphylococci); no fungal infections were detected. Three cancer patients and 2 organ transplant recipients had respiratory viral infections diagnosed by symptoms and positive direct fluorescent antibody test results. Of the 39 children, levels of liver enzymes were obtained for 28 as part of routine care. Of these children, 6 had abnormal values; 2 of these 6 also had virus detected (HHV-6B in 1 patient and influenza A in the other). The other 4 patients with increased levels of liver enzymes had longstanding abnormalities; review of their records suggested that the abnormal results were caused by underlying disease in 1 liver transplant patient and chemotherapy in 3 cancer patients. Six patients have died since completion of our study, none from infectious causes.

Of the 11 specimens tested in the shell vial assay, none was positive for CMV. CMV DNA was detected by PCR in lymphocytes of 1 (3.3%) of 30 cancer patients and 3 (33%) of 9 organ transplant recipients (Table 2).

Lymphocyte cultures for HHV-6 and HHV-7 were obtained for 34 of the 39 children. Of these, 15 cultures showed cytopathic effects similar to those commonly reported for HHV-6, but none of these were positive for HHV-6 by PCR. Cultured cells from 1 patient with leukemia (study participant 19) were positive for HHV-6B by PCR, in the absence of a cytopathic effect. Four (13%) of 30 cancer patients and 3 (33%) of 9 transplant recipients (p = 0.19) had HHV-6B DNAemia; HHV-6A was not detected. Viral loads ranged from 50 to 500,000 DNA copies/mL. Of 7 PCR-positive patients, 3 also showed positive results in peripheral blood mononuclear cells (PBMC) and whole blood specimens, 2 in only PBMC, and 1 only in a lymphocyte culture. Patient 40, a transplant recipient, had positive results in all specimens tested including PBMC, whole blood, and plasma, with the highest viral load (>5 × 10⁵ genomes/mL) in PBMC. These results are consistent with either severe acute infection or the child having germline integrated HHV-6B (32,33). PCR positivity was more common in children whose blood was collected within 6 months of initiation of immune suppression compared with children sampled >6 months after initiation of immune suppression (cancer patients: 4 of 16 vs. 0 of 14 after 6 months, p = 0.050; transplant recipients: 2 of 3 vs. 1 of 6, p = 0.26) (Figure). Solid-tumor patients were less likely to be positive for HHV-6B than either leukemic (1 of 20 vs. 3 of 9, p = 0.046) or SOT patients (1 of 20 vs. 3 of 9, p = 0.046). Two of the HHV-6B-positive SOT patients had concurrent CMV DNAemia.

HHV-7 DNA was not detected in any patients. Lymphocyte cultures were not tested for HHV-7 by PCR because none of the plasma, whole blood, or leukocyte pellets were positive for the virus.

Patients who were positive for β-herpesviruses and those who were negative for these viruses did not differ in their average fever temperature (38.5°C vs. 38.6°C). In SOT recipients, the prevalence of HHV-6B and CMV and the spectrum of their associated illnesses were concordant with prior reports. Thus, we will focus on the cancer patients.

The only cancer patient with CMV DNAemia appeared to have primary CMV infection on the basis of negative results for immunoglobulin (Ig) G to CMV, negative bacterial blood culture, and age (online Appendix Table, available from www.cdc.gov/EID/content/14/4/579-appT.htm). This patient was a 5.4-year-old child (study participant 29) with acute lymphoblastic leukemia (ALL) in remission; he had upper respiratory symptoms and fever but otherwise appeared well. He was treated with an antimicrobial drug (ceftriaxone) on an outpatient basis, but his blood culture was negative for bacteria and no other source for his fever and illness was found. This child began cancer chemotherapy 36 months before the febrile event.

Four patients appeared to have reactivated HHV-6B infection on the basis of positive results for virus DNA, high-avidity IgG, and clinical symptoms. The first patient (study participant 19) was a 2-year-old boy with ALL in remission for the first time. In the second month of chemotherapy; he was hospitalized for fever and ill appearance, but he was not neutropenic or lymphopenic. He had marked hepatomegaly (≈10 cm below the right costal margin) and abnormal results in liver function tests. He was receiving fluconazole for oral thrush, and his liver enzyme levels were mildly increased before the febrile illness. However, at admission he had an aspartate aminotransferase (AST) level of 2,356 U/L and an alanine aminotransferase (ALT) level of 2,410 U/L. HHV-6 and HHV-7 IgM and IgG titers were obtained (Specialty Laboratories, Valencia, CA, USA) as part of his ALL treatment protocol. His HHV-6 IgM titer was negative (<40), and his IgG titer was positive (320) and high-avidity antibodies were detected. Results for HHV-7 IgM were negative, and those for IgG were weakly positive (titer 20). Other studies conducted by the primary medical team included tests for acute hepatitis and a CMV PCR; all results were negative. Four days later, his AST level was 84 U/L and his ALT level was 314 U/L. On follow up 7 days after admission, both values were near normal. The hepatitis was attributed to methotrexate, which was therefore withheld until the levels of liver enzymes returned to normal. On the basis of the positive culture for HHV-6B and hepatitis in the setting of febrile illness without other identified infections, we suspect that reactivated HHV-6B infection caused the fever and exacerbated the hepatitis.

A 5-year-old girl with relapsed acute myeloid leukemia (study participant 23) was intermittently febrile while hospitalized for 1 month during chemotherapy. She was treated with broad-spectrum antibacterial drugs and antifungal drugs. At study enrollment on day 12 of hospitalization, she had an absolute neutrophil count of 0 and an absolute lymphocyte count of 560 cells/μL. Daily blood cultures and sporadic respiratory viral studies did not reveal a cause for her fever, but her whole blood and PBMC were positive for HHV-6B by PCR.

HHV-6B was detected in a 2-year-old girl with ALL in remission (study participant 25) and in a 22-month-old boy with stage 3 Wilms tumor (study participant 38). Both of these patients were lymphopenic and had nonspecific febrile illnesses; neither had other identified sources of fever.

We suspect primary CMV infection in 1 patient and reactivated HHV-6B infection in 4 cancer patients as likely causes of the febrile episodes. This conclusion is based on a combination of clinical data, positive PCR findings, and lack of other identified sources for fever.