Routine Chest Radiography for the Evaluation of FebrileNeutropenic Patients After Autologous Stem

Cell Transplantation

Vivek Roy,* Lubna I. Ali, and George B. SelbyDepartment of Medicine, Hematology-Oncology Section, University of Oklahoma Health Science Center, Oklahoma City, Oklahoma

Chest radiographs are routinely obtained for diagnostic evaluation of neutropenic febrilepatients. We investigated the frequency of chest radiographic abnormalities during fe-brile episodes after autologous PBSC transplants and assessed the relationship of theseabnormalities to past history of pulmonary disease, pre-transplant chest radiographicabnormalities, and pulmonary signs or symptoms at time of fever. We also studied theimpact of chest radiographic findings on patient management. Sixty-one consecutiveadult autologous PBSC transplant recipients were studied. Fifty-two (85%) developedfever, and 20 (38%) of these showed new chest radiographic abnormalities suggestive ofpulmonary infection. Patients with pre-transplant chest radiographic abnormalities weremore likely to develop additional abnormalities with fever post-transplant. Pulmonarysymptoms or signs had low sensitivity or specificity for predicting radiographic abnor-malities. Only 40% of patients with pulmonary symptoms or signs had an abnormal chestradiograph. Twenty-six percent of patients with abnormal chest radiographs had no clini-cal findings suggestive of pulmonary infection. The identification of chest radiographicabnormality did not change empiric antibiotic treatment in any patient. The role of routinechest radiography for diagnostic evaluation of febrile autologous PBSC transplant pa-tients should be re-evaluated. Am. J. Hematol. 64:170–174, 2000. © 2000 Wiley-Liss, Inc.

Key words: BMT; pulmonary infection; febrile neutropenia; chest radiographs

INTRODUCTION

Neutropenic fever is a common and potentially life-threatening problem in patients receiving chemotherapy[1–3]. Infection often appears in the lungs in a largeproportion of these patients [4,5]. Non-infectious pulmo-nary processes such as alveolar hemorrhage, drug toxic-ity, or ARDS may also result in fever in chemotherapyrecipients [6]. A chest radiograph is commonly recom-mended in the diagnostic evaluation of febrile neutrope-nic patients [7]. This recommendation is based on theobservation that a significant proportion of febrile neu-tropenic patients with pneumonia may have a normalphysical examination [2,8] and on the assumption thatidentification of abnormal chest radiograph will improvepatient management. The utility of this practice has beenquestioned in recent years. Several authors have sug-gested that chest radiography does not contribute to themanagement of patients who develop febrile neutropeniafollowing conventional chemotherapy [9,10]. Bone mar-

row or PBSC transplant recipients generally have ahigher degree of immunosuppression and a different fre-quency and nature of pulmonary problems than patientsreceiving conventional chemotherapy. The role of chestradiography in the initial management of febrile post-transplant patients has not been critically evaluated. Wesuspected that the yield of routine chest radiography forinitial diagnostic evaluation and management of post-transplant febrile neutropenic patients is low. To addressthis issue, we conducted a retrospective study to assessthe frequency of abnormal chest radiographs, the asso-ciation of abnormal chest radiographs with clinical pic-ture, and the influence of chest radiography findings on

*Correspondence to: Vivek Roy, M.D., Department of Medicine,Heme-Onc Section, WP 2010, Box 26901, Oklahoma City, OK 73190.E-mail: vivek-roy@ouhsc.edu

Received for publication 25 June 1999; Accepted 2 February 2000

American Journal of Hematology 64:170–174 (2000)

© 2000 Wiley-Liss, Inc.

the initial empiric treatment of the febrile episode inautologous peripheral blood stem cell transplant recipi-ents.

MATERIALS AND METHODS

We reviewed the records of 61 consecutive adult pa-tients with breast cancer (N 4 33), non-Hodgkin’s lym-phoma (N 4 20), or Hodgkin’s disease (N 4 8) whoreceived high-dose chemotherapy with autologous PBSCtransplantation at the University of Oklahoma Marrowand Blood Transplantation Program between January1996 and January 1998. Data related to any febrile epi-sodes during their transplant course was retrieved. Feverwas defined as oral or tympanic membrane temperaturemore than 38.5°C lasting at least 2 hr. Any fever two ormore days after defervescence of the first fever was con-sidered to be a second fever. All patients had a chestradiograph at the time of first fever episode. Radiographswere reviewed by a staff radiologist and compared tobaseline (pre-transplant) films when available.

All patients received ofloxacin, acyclovir, and flu-conazole for infection prophylaxis. At the onset of neu-tropenic fever, a detailed history and physical examina-tion was performed and empiric antibiotic therapy withintravenous vancomycin and ceftazidime (aztreonam incephalosporin-allergic patients) was started after appro-priate cultures (blood, urine, stool, and sputum) had beentaken. Treatment with amphotericin B was started as pre-sumptive therapy for fungal infection if fever persistedfor more than 48–72 hr in the absence of an etiologicdiagnosis. Additional chest radiographs or other imagingstudies were obtained at the discretion of the treatingphysician. Patients with second or subsequent febrile epi-sodes were treated at the discretion of the attending phy-sician.

Information about the clinical symptoms, physicalfindings, complete blood counts, chest radiographic find-ings at baseline and at the time of febrile episode, andmicrobiological culture results was collected by a retro-

spective chart review. Data related to the response toinitial antibiotic therapy, any changes in antibiotictherapy, the documented reason for the change in antibi-otic therapy, and etiology of fever were also obtained bychart review.

Data was analyzed to assess the relationship betweenclinical status and radiological findings and whetherchest radiography led to an etiologic diagnosis or re-sulted in change in empirical therapy. Statistical signifi-cance of the findings was determined using two-tailedFisher’s Exact Test. Stepwise logistical regression analy-sis was used to assess the sign or symptom most predic-tive of a chest radiographic abnormality.

RESULTS

Sixty-one patients were studied. They ranged in agefrom 19 to 68 years (median 48). Thirty patients had ahistory of chronic cigarette smoking. Seventeen patientshad a history of pulmonary disease or had pulmonarysymptoms prior to transplantation. Eleven of these pa-tients developed fever during the transplant course. Pre-transplant chest radiographs were available in 56 patientsand abnormal in 18 patients, 14 of whom developed fe-ver. Pre-transplant chest radiographic abnormalities andadditional abnormalities seen with fever are shown inTable I.

Fifty-two patients developed fever during their trans-plant course. Forty-nine patients were neutropenic (ANC< 100) at the time of fever. Three patients had absoluteneutrophil count (ANC) between 500 and 2,800/ml at thetime of initial rise of fever, but because of rapidly dete-riorating neutrophil count, they were treated as neutro-penic. The average duration of neutropenia after trans-plantation was 13 days. The median day of first fever wasday +3 with the day of PBSC infusion called day 0 (rangeday −7 to day +12). The median duration of fever was 5days (range 1 to 16 days). Thirty-five patients had signsor symptoms suggestive of pulmonary disease at the timeof fever; 14 patients had two or more symptoms or signs.

TABLE I. Pre-transplant Chest Radiographic Abnormalities and Additional Abnormalities at the Timeof Fever

Pre-transplant radiologicalabnormalities (N 4 18)

Fever(N 4 14)

No additional chestradiographic abnormality

with fever (N 4 4)

Additional chest radiographicabnormalities with fever (N 4 10)

Atelectasis/infiltrates Consolidation

Calcific granulomas (3) 3 2 1Pleural effusions (4) 2 2Pulmonary nodules (1) 1 1Prominent pulmonary vessels (2) 1 1Chronic interstitial fibrosis (2) 2 2Cephalization of blood flow (1) 1 1Pleural thickening (2) 2 2Pulmonary infiltrates (2) 2 1 1Emphysema (1)

Chest X-Rays for Post-Transplant Neutropenic Fever 171

These findings and the associated chest radiographic ab-normalities are shown in Table II. All 52 febrile patientshad a chest radiograph with their first fever. Seventeenpatients developed new abnormalities not seen on theirbaseline chest radiographs. Second or subsequent chestradiographs were obtained in 29 patients (16 with normalfirst chest radiograph) who remained febrile for morethan 48 hr or developed respiratory symptoms or signs.Only three new chest radiograph abnormalities werenoted among these patients.

Radiographic abnormalities included infiltrates or at-electasis (9), consolidation (10), and pulmonary edema(1). The most common abnormality was consolidationsuggestive of pneumonia. Smokers tended to have base-line chest radiographic abnormalities and develop newabnormalities with fever more frequently than nonsmok-ers, although this did not reach statistical significance (P4 0.1). Of the 24 smokers who became febrile, 9 (37%)had abnormal baseline chest radiographs, and 6 devel-oped additional abnormalities with fever. Of the 28 non-smokers who became febrile, 5 (17%) had abnormalbaseline chest radiographs, and 4 of these developed ad-ditional abnormalities with fever. Table I shows the chestradiographic abnormalities in febrile patients.

Overall, 18 patients had abnormal baseline chest ra-diographs, and 14 of these developed fever. Ten of these14 (71%) patients had additional abnormalities with thefever. In contrast, only 10 (26%) of 38 febrile patientswith normal baseline chest radiographs developed anyabnormality with fever (P 4 0.03). Records of 51 pa-tients were available to evaluate for signs and symptomsat the time of fever (1 chart was lost). Of the 35 patientswho had pulmonary signs or symptoms at the time offever, 14 had abnormal chest radiograph (40%). Five of16 (31%) patients without any sign or symptom of pul-monary disease at the time of fever had abnormal chestradiograph. In a multivariate analysis, the symptom orphysical finding most likely to be associated with anabnormal chest radiograph was cough and localized areaof decreased breath sounds on auscultation (P 4 0.06).Fourteen of 19 (73%) patients with abnormal chest ra-

diographs had at least one symptom or sign suggestive ofpulmonary infection (Table II).

The cause of fever could be determined in 21 of 52patients (40%). Another 6 patients had amphotericin B-responsive fever suggesting a fungal infection, but de-finitive diagnosis could not be made (Table III). In 16(76%) of these patients, the diagnosis was based on chestradiographs showing infiltrates or consolidation sugges-tive of pneumonia. Although chest radiographs helpedlocalize the infection, empiric antibiotic therapy was notchanged in any patient on the basis of radiographic find-ings. Antibiotic therapy was changed in 5 (9.6%) patientswith continuing fever, and in 3 of these there were ad-ditional chest radiographic abnormalities. The choice ofantibiotics was influenced by chest radiographic findingin only one of the three patients in whom interstitialinfiltrates suggestive of Mycoplasma pneumonia werenoted. In the other two patients, antibiotic choice wasbased on demonstration ofClostridium difficilecytotoxinin stool (1 patient) and identification ofAspergillussp. ina cutaneous lesion (1 patient).

DISCUSSION

We investigated the utility of chest radiography in thediagnostic evaluation of patients with neutropenia andfever after PBSC transplantation. We assessed the fre-quency of chest radiographic abnormality, the associa-

TABLE III. Causes of Fever in Febrile Neutropenic PBSCTransplant Recipients*

Diagnosis Number of patients (%)

Pneumonia 13 (30.7%)Sinusitis 1 (1.9%)Bacteremia 3 (5.7%)CVC infection 1 (1.9%)Amphotericin-responsive fever 6 (11.5%)Unknown 26 (50%)

*Definitive diagnosis could be made in 21 of 52 patients. CVC, centralvenous catheter; PBSC, peripheral blood stem cell.

TABLE II. Clinical Symptoms and Signs at the Time of Fever and the Associated ChestRadiographic Findings

Sign or symptom (N)Number with

normal radiographs

Number withradiographicabnormalities

Radiographic abnormality

Atelectasis/infiltrate Consolidation CHF

Tachypnea (19) 14 5 2 3Decreased breath sounds (12) 5 7 3 3 1Cough (11) 4 7 1 6Rales or crackles (4) 1 3 1 2Dyspnea (3) 1 2 1 1Sputum (3) 1 2 2Wheezing or rhonchi (3) 1 2 2Chest pain (2) 2 0

172 Roy et al.

tion of abnormal radiographs with the clinical picture,and the effect of chest radiographic findings on the initialempiric treatment of fever. Fifty-two of 61 (85%) pa-tients developed fever during the transplant course, and20 of these had abnormal chest radiographs (38%).Thirty-five febrile patients had symptoms or signs sug-gestive of respiratory infection, and 14 of these (40%)had abnormal chest radiographs. Presence of an abnor-mality on pre-transplant chest radiograph was stronglyassociated with development of additional abnormalitiesduring fever (P 4 0.03). Chest radiographs obtained atthe time of first fever did not lead to a change in initialempiric antibiotic therapy in any patient.

Our finding of a 38% incidence of chest radiographicabnormalities is somewhat higher than that reported byother investigators. Incidence of chest radiographic ab-normalities ranging from 6% to 22% has been reported infebrile neutropenic patients after conventional-dose che-motherapy [8,9,11,12]. This difference may be related todifferences in patient populations, patient selection, ordegree of immune suppression. The reported studieslargely deal with patients who were neutropenic afterconventional chemotherapy. Our study dealt only withadults, who had previously received multiple cycles ofchemotherapy for their malignancy and were now neu-tropenic after high-dose chemotherapy and PBSC trans-plantation. Nearly half of our patients were chronic ciga-rette smokers, and about a third had abnormalities on abaseline chest radiograph, factors that may explain thehigher incidence of chest radiographic abnormalitieswhen these patients developed neutropenia and fever.Our patients had severe neutropenia (median ANC at thetime of fever < 100 × 106/l) lasting for a median of 13days. The degree and duration of neutropenia were notreported in other studies.

73% of patients with abnormal chest radiographs inour series had signs or symptoms suggestive of pulmo-nary disease, but the specificity of physical examinationwas low. Only 40% of patients with pulmonary signs orsymptoms at the time of neutropenic fever in our serieshad abnormal chest radiographs. Other investigators havealso commented on this lack of specificity of physicalexamination findings in both neutropenic [12] and non-neutropenic [13] patients. Donowitz et al. [12] reportedthat 30% febrile neutropenic patients with some signs orsymptoms of pulmonary disease had a normal chest ra-diograph. The predictive value of normal physical ex-amination in excluding a pulmonary abnormality hasbeen debated in the literature. In a study by Feusner et al.[11] only 1 of 49 (2%) febrile patients without pulmonarysigns or symptoms had an abnormal chest radiograph.Jochelson et al. [14] also found a similarly low incidence(4%) of finding pulmonary infection by chest radiogra-phy in febrile neutropenic patients without pulmonarysigns or symptoms. In our series 30% of febrile neutro-

penic patients without pulmonary signs or symptoms hada chest radiographic abnormality suggestive of pneumo-nia. This is similar to the findings reported by other in-vestigators [2,12] who found 20–25% incidence of pneu-monia in febrile neutropenic patients without pulmonarysigns or symptoms. It is to diagnose this group of patientsthat obtaining a routine chest radiograph during diagnos-tic evaluation of febrile neutropenic patients has beenrecommended.

The importance of making a diagnosis of pulmonaryinfection has to be viewed in the context of the overallimpact on patient management. The identification of achest radiographic abnormality in the absence of symp-toms or signs is of value if it leads to better patientmanagement. In our series, the initial empiric therapywas not affected by finding a radiographic abnormality inany patient. Lack of effect of chest radiographic findingson the initial management of patients who were neutro-penic and febrile after conventional chemotherapy haspreviously been reported in the literature [9,12,14]. Wenow show a similar lack of impact of chest radiographicfindings on the initial treatment of post-transplant febrileneutropenic patients. Current management of febrile neu-tropenic patients involves empiric use of potent, high-dose, broad-spectrum, intravenous antibiotics that are ef-fective against most of the commonly encounteredpathogens. Anatomic localization of infection is there-fore not likely to lead to a change in antibiotic therapy.

Our study supports the findings in the literature thatchest radiography can identify pulmonary infection in asignificant number of febrile neutropenic patients inwhom clinical findings are unrevealing. Chest radiogra-phy is more likely to be informative if the patient has aknown baseline chest radiograph abnormality or hasclinical findings suggestive of a pulmonary infection.However, given the current practice of treating febrileneutropenic episodes with high-dose, broad-spectrum an-tibiotics, the identification of lungs as the site of infectionis unlikely to influence patient management. We provideevidence in favor of the proposition that, like with con-ventional chemotherapy, routine chest radiograph is notnecessary in the initial management of febrile neutrope-nic patients after PBSC transplantation. However, if fe-ver persists or additional physical findings develop, achest radiograph may contribute to patient managementby revealing a pattern of abnormality suggestive of aninfection not adequately treated by the initially chosenantibiotics, or be helpful in directing further diagnosticefforts toward or away from the lungs. The number ofpatients in our study was not sufficient to allow us toevaluate the role of subsequent chest radiographs in pa-tients with a normal initial chest radiograph who con-tinue to remain febrile after initial antibiotic treatment.

We emphasize that our patients included autologousPBSC transplant recipients only. Allogeneic transplant

Chest X-Rays for Post-Transplant Neutropenic Fever 173

recipients constitute an entirely different patient popula-tion because of more severe and longer duration of im-munosuppression and higher likelihood of non-infectiouspulmonary complications than PBSC transplant or con-ventional chemotherapy recipients. The findings reportedhere may not be applicable to allogeneic transplant re-cipients. Studies to evaluate the role of routine chestradiographs in the initial work-up of fever after alloge-neic transplantation is currently in progress.

REFERENCES

1. Browder AA, Hoff J, Petersdorf RG. The significance of fever inneoplastic disease. Ann Intern Med 1961;55:932–942.

2. Sickles EA, Greene WH, Wienik PH. Clinical presentation of infection infebrile granulocytopenic patients. Arch Intern Med 1975;135:715–719.

3. Sable CA, Donowitz GR. Infections in bone marrow transplant recipi-ents. Clin Infect Dis 1994;18:273–284.

4. Singer C, Kaplan M, Armstrong D. Bacteremia and fungemia compli-cating neoplastic disease. Am J Med 1979;62:731–742.

5. Pizzo PA, Robichaud KJ, Gill FA, Witebsky FG. Empiric antibioticand antifungal therapy for cancer patients with prolonged fever andgranulocytopenia. Am J Med 1982;72:101–111.

6. Rosenow EC, Wilson WR, Cockerill FI. Pulmonary disease in theimmunocompromised host. Mayo Clinic Proc 1985;60:610–631.

7. Pizzo PA, Meyers J, Freifeld AG, Walsh T. Infections in the cancerpatient. In: DeVita VT, Hellman S, Rosenberg SA, editors. Cancer:principles and practice of oncology, 4th edition. Philadelphia: JB Lip-pincott Company; 1993. p 2292–2337.

8. Sickles EA, Young VM, Greene WH, Wiernik PH. Pneumonia in acuteleukemia. Ann Intern Med 1973;79:528–534.

9. Katz JA, Bash R, Rollins N, Cash J, Buchanan GR. The yield ofroutine chest radiography in children with cancer hospitalized for feverand neutropenia. Cancer 1991;68:940–943.

10. Korones DA, Hussong MR, Gullace MA. Routine chest radiography ofchildren with cancer hospitalized for fever and neutropenia. Is it reallynecessary. Cancer 1997;80:1160–1164.

11. Feusner J, Cohen R, O’Leary M, Beach B. Use of routine chest radi-ography in the evaluation of fever in neutropenic pediatric oncologypatients. J Clin Oncol 1988;6:1699–1702.

12. Donowitz GR, Harman C, Pope T, Marc Stewart F. The role of thechest roentgenogram in febrile neutropenic patients. Arch Intern Med1991;151:701–704.

13. Heckerling PS. The need for chest roentgenograms in adults with acuterespiratory illness. Arch Intern Med 1986;146:1321–1324.

14. Jochelson MS, Tsculer J, Stomper PC. The yield of chest radiographyin febrile and neutropenic patients. Ann Intern Med 1986;105:708–709.

174 Roy et al.