Infection in the Myelodysplastic Syndromes CLAIRE POMEROY, M.D., MARTIN M. OKEN, M.D., ROBERT E. RYDELL, M.D., GREGORY A. FILICE, M.D.,

Minneapolis, Minnesota

PURPOSE: To determine the incidence, characteristics, and o u t c o m e of infection in patients with myelodysplastic syndromes (MDS) and risk factors that may lead to infection. PATIENTS AND METHODS: W e reviewed infections that occurred in 86 consecutive patients with M:DS who received care f r o m 1968 to 1986 at a university-affiliated Veternn~ Affairs Medical Center. Time lines charting the course of each patient with MDS were created and included infections, MDS subgroup at the time of presentation and at the time of each infection, peripheral neutrophil counts, and therapies for MDS. RESULTS: Infections occurred at a rate of nearly one per patient year of observation. Infection rates were associated with MDS subgroup as follows: refractory anemia with or w i t h o u t ringed sideroblasts (RA ± RS) less than refractory anemia with excess blasts (RAEB) less than ~RAEB in transformation (RAEB-T). The g r o u p of RA ± RS patients who had erythroid abnormalities but minimal or no dyspoiesis of other

cell lines had the lowest rate of infections. Infection rates were higher in patients with less than or equal to 1,000 neutrophilA/#L blood than in patients with greater than 1,000 neutrophilA/#L blood for each classifiable MDS subgroup. Neutrophil c o n c e n t r a t i o n and M D S subgroup were independent risk factors for infection in patients with MDS. Bacterial pneumoniA~ and skin abscesses were the most c o m m o n infections. Infection was the most c o m m o n cause of death during MDS, accounting for 64% of deaths, and was m o r e c o m m o n than transformation to acute leukemla as a cause of death. CONCLUSION: Infection is a c o m m o n , l i f e -

threatening problem in patients with MDS. Neutropenia and MDS subgroup are each risk facFrom the Infectious DiseaseSection (CP, GAF) and HematologySection (MMO), Medical and Research Services(CP, MMO, GAF) and the Pathology Service (RER), VeteransAffairs Medical Center, and the Departments of Medicine(CP, MMO,GAF)and Pathology(RER), University of Minnesota, Minneapolis,Minnesota. Requests for reprints should be addressedto Claire Pomeroy, M.D., MinneapolisVeteransAffairs Medical Center (151), One Veterans Drive, Minneapolis, Minnesota55417. Manuscript submitted March 30, 1990, and accepted in revised form October 1, 1990.

338

March 1991

The American Journal of Medicine

tors for infection. Clinicians should aggressively evaluate patients with fever and MDS for infection, especially pneumonia and skin infections.

Yelodysplastic syndromes (MDS) are a heterogeneous group of clonal disorders of the M bone marrow, manifested by refractory cytopenias and characteristic morphologic abnormalities of the peripheral blood and bone marrow. The prevalence of MDS may be as high as one in 500 in people over the age of 55 years [1]. Older names for this syndrome included preleukemia and smoldering leukemia, which contributed to the erroneous impression that progression to acute leukemia is the major cause of death. The French, American, British (FAB) classification system facilitates systematic analysis of these disorders [2]. Although transformation to acute leukemia has been reported with variable frequency [3,4], previous reports have suggested that complications of bone marrow failure, including infections and hemorrhage, are the major causes of death [3,5]. However, detailed studies of infectious complications have not been reported. We reviewed records of 86 consecutive patients with MDS and characterized infectious complications and the contributions of risk factors, including MDS subgroup, neutropenia, and immunosuppressive therapy.

PATIENTS AND METHODS We studied patients with MDS at the Minneapolis Veterans Affairs Medical Center during the period 1968 to 1986. The hospital cancer registry, the Hemato]ogy/Oncology Service patient log, and the Bone Marrow Laboratory log were reviewed to identify patients suspected of having MDS. Since several classification schemes were used over this time period, a variety of diagnoses that might have represented MDS were used to identify possible patients, including "preleukemia," "subacute leukemia," "smoldering leukemia," "atypical chronic myelocytic leukemia," "sideroblastic anemia," "refractory anemia," and "myelodysplasia." Peripheral blood and bone marrow samples were reviewed, and patients were included if cytopenias and dysmyelopoiesis were sufficient to meet diagnostic criteria for MDS [2]. Patients were excluded if they

Volume 90

INFECTION IN THE MYELODYSPLASTIC SYNDROMES / POMEROY ET AL

presented with acute leukemia with greater than 30% marrow blasts or if they had nutritional deficiency or aplastic anemia. Patients were considered to have secondary MDS if their disease occurred after chemotherapy or radiation therapy for another condition. Patients with simultaneous malignancies that might have contributed to their risk of infection were also identified. MDS was categorized by FAB subgroup at presentation of MDS and whenever the patient met criteria for progression to a more severe MDS subgroup [2]. Briefly, refractory anemia (RA) and RA with ringed sideroblasts (RARS) were limited to cases with less than 1% blasts in peripheral blood and less than 5% blasts in bone marrow. RA with excess blasts (RAEB) included cases of RA with granulocytic immaturity but less than 5% blasts in peripheral blood and 5% to 20% blasts in bone marrow. RAEB in transformation (RAEB-T) included cases with greater than 5% blasts in peripheral blood and 20% to 30% blasts in bone marrow or the presence of Auer rods. Chronid myelomonocytic leukemia (CMML) included cases of RAEB with greater than 109 monocytes per liter of peripheral blood. Patients with RA and RARS were combined and referred to as RA ± RS. For analysis, we divided RA ± RS patients into two groups: those with erythroid abnormalities but minimal or no dyspoiesis of other cell lines (group I) and those with multilineage abnormalities (group II). All available bone marrow samples were reviewed, and each patient's F A B s u b g r o u p was u p d a t e d w h e n e v e r M D S evolved to a different classification. Infection risks were analyzed in two ways: by MDS subgroup at presentation of MDS and at the time of each infection. Time lines were created for each patient indicating the MDS subgroup, peripheral neutrophil counts, and immunosuppressive therapies received during each month of illness. Patient years of observation in each category were summed. Medical records were reviewed for demographics, clinical presentation of MDS, progression and treatment of MDS, infectious complications, noninfectious complications, and time and cause of death. We limited our analysis to infections requiring hospitalization or occurring during hospitalization. Our ability to ascertain infections when patients were outside the hospital was limited by the fact that some patients lived far away and/or received care for minor problems outside of our hospital. Criteria for diagnoses of infectious diseases were as follows: For pneumonia, patients had to have fever and cough, treating physician assessment of pneumonia, and infiltrates on chest roentgenogram. Patients with fever and cough but without

RA * RSGroup I

RA *_.RSGrOUp II

RAEB

RAEB-T

CMML

Unclass

AML/ALL

3-

Figure 1. Course of the myelodysplastic syndrome in the 86 patients studied. Progression of disease is read from left to right. Dots in each diagnostic column indicate that patients met criteria for that FAB subgroup or for transformation to acute leul~emia at some point during their illness. The number of patients observed with each pattern of progression is indicated by the numbers under each circle. Numbers at the bottom of each column indicate the total number of patients analyzed in each FAB subgroup at any point during the course of their illness.

infiltrates were considered to have tracheobronchitis. For urinary tract infection, patients had to have pyuria and either greater than or equal to 105 colony-forming units of bacteria per milliliter of urine or bacteria seen on urine gram stain, as well as treating physician impression of urinary tract infection. For bacteremia, patients had to have at least one blood culture yielding an organism considered a pathogen by the treating physician. For sinusitis, patients had to have an air-fluid level or mucosal thickening on sinus roentgenograms and treating physician impression of sinusitis. For Clostridium difficile-associated colitis, patients had to have loose or watery stools and identification of C. difficile toxin in a stool specimen or pseudomembranes observed by sigmoidoscopy. For meningitis, patients had to have a positive cerebrospinal fluid (CSF) culture, greater than five white blood cells in the CSF, and treating physician impression of meningitis. We accepted treating physician clinical diagnoses of cellulitis, perianal abscess, oral herpes

March 1991 The American Journal of Medicine Volume 90

339

INFECTION IN THE MYELODYSPLASTIC SYNDROMES ' POMEROY ET AL TABLE I Infectious Diseases in MDS Patients FAB Subl~roupat Onsetof Infection RA + RS Group I (70.4)t [1915

Group II (95.7) [51]

RAEB (11.2) [19]

RAEB-T (6.5) [17]

Other* (16.3) [6]

Pneumonia Skin infection Urinary tract infection Bacteremia Thrush

6 3 4

18 12 13

8 6 3

10 3 4

2 10 1

44 34 25

23 18 13

2 0

5 2

7 1

5 3

3 4

22 10

12 5

Bronchitis C. difficile-

2

3

2

1

0

8

4

0 0 2

4 4 3

2 1 0

1 1 1

0 O 0

7 6 6

4 3 3

2

1

1

1

0

5

3

O 0

3 1

0 1

0 1

0 0

3 3

2 2

6~ 75

1~ 33

5 *~ 36

0 20

15 188

8 100

associated colitis Sinusitis Perirectal abscess Abdominal abscess§ Epi-prchitis Herpes labialis Miscellaneous Total

3 24

Total (200.1) [86]

Percent of Infections

° CMML and unclassified MDS. T( ) = patient years of observation Ior each FAB subgroup. [ ] = number of patients observedin each MDS subgroup at some point during the course of their illnesses. Manypatients spenl time in more than one FAB subgroup. § One caseof ruptured appendixwith peritonitis and abscess, two casesof diverticular abscess,one caseof postoperativewound infection with pelvic and intra-abdominal abscess, and one case of surgical wound infection. One case each of endocarditis, bacterial thyroiditis, and vocal cord inflammation with granulomas. ' One case each of blepharitis, miliary tuberculosis, prostatitis, parotitis, Epstein-Barrvirus hepatitis, and myobomitis. # One disseminated aspergillusinfection. * * Two casesof intravenouscatheter-associatedphlebitis, and one case each ot disseminatedcandida infection, salivarygland abscess, and mastoiditis.

,TABLE II Infection Rates in Patients with MDS FAB Subgroupat Onsetof Infection RA ± RS Group I

Group II

RAEB

RAEB-T

Other*

Total

All MDS

0.34t (70.4)~ [19]§

0.78 (95.7) [51]

2.96 (11.2) [19]

5.54 (6.5) [17]

1.22 (16.3) [6]

0.94 (200.1) [86]

MDS, excluding secondary MDS

0.26 (65.6) [18]

0.77 (92.2) [48]

3.05 (9.2) [17]

5.54 (6.5) [17]

1.01 (15.8) [5]

0.89 (189.3) [80]

MDS, excluding simultaneous malignancy

0.26 (65.6) [18]

0.84 (83.3) [47]

2.96 (11.2) [19]

5.54 (6.5) [17]

1.01 (15.8) [5]

0.94 (182.4) [80]

' CMMLand unclassified MDS. r Infections per patient year. ( ) = patient years of observation tor each FABsubgroup. [ ] = number of patients observedin each MDS subgroup at some point during the course of their illnesses. Many patients spent time in more than one category.

simplex infection, and oral candidiasis. We accepted diagnoses of intra-abdominal and pelvic abscesses if they were observed at surgery. We also noted unexplained febrile illnesses with an oral temperature greater than or equal to 38.3°C (101°F) documented in the hospital that could not be attributed to other causes, such as infection or a reaction to blood products or a drug. For inclusion in this category, patients had to have had an evaluation including a history and physical examination, blood and urine cultures, and chest roentgeno340

March 1991 The American Journal of Medicine Volume 90

grams. Causes of death were ascertained based on treating physician assessments, autopsy results, and death certificates. Survival curves and probabilities were calculated by the actuarial method, p values for comparisons of subgroups were calculated for time to event by the Lee-Desu statistics (Statistical Packages for the Social Sciences, Version 2.2, SPSS-X, SPSS Inc., Chicago, Illinois). p values were determined by the generalized Wilcoxon (Breslow) and generalized Savage (Mantel-Cox) methods.

INFECTION IN THE MYELODYSPLASTIC SYNDROMES / POMEROY ET AL TABLE III Infection Rates In Patients with MDS According to Blood Neutrophil Concentrations at Time of Infection FAB Subgroupat Onsetof Infection Blood Neutrophll Concentration (number//AL)