Correspondence
Clinical Problem-Solving: Hypereosinophilic Syndrome
N Engl J Med 1996; 334:538-540February 22, 1996
- Article
To the Editor:
The Clinical Problem-Solving article (Nov. 2 issue)1 describing a patient with hypereosinophilic syndrome who had a particularly comprehensive series of diagnostic and ultimately misleading or fruitless tests provides an excellent example of one of the major problems in medicine today, that of excessive diagnostic testing and the economic consequences. We believe the case report is particularly useful and represents almost a paradigm of the problem. In their commentary, Drs. Putterman and Ben-Chetrit contrast the “liberal tester” (in this case the primary attending physician) and the “parsimonious tester” (the discussant) and highlight the reasoning and appropriateness of the decisions of the parsimonious tester while at the same time admonishing one not to “throw stones” at the liberal tester, because the choice “was not an easy one” and the focus should be on identifying the causes of excessive testing in order to correct the situation or provide more cost-effective guidelines.
Putterman and Ben-Chetrit list an array of causes of excessive testing but in the case under discussion, I speculate that the primary attending physician held the liberal-testing stance out of ignorance. Virtually all the diagnostic tests were directed at detecting a malignant cause of the clinical syndrome. However, cancer should have been of the least concern to the primary physician for the following reasons: the hypereosinophilic syndrome is vastly more frequent than an eosinophilic response to an occult carcinoma; an occult carcinoma that would cause hypereosinophilia is more than likely to be incurable and untreatable; and most important, a trial of corticosteroids is not contraindicated by virtually any of the diagnostic possibilities that one could consider (including cancer) and, as indicated by the authors, could reasonably be considered the most economical diagnostic test of all.
In my mind, there is no question that a trial of therapeutic cortisone was the diagnostic test and treatment of choice and should have preceded any invasive procedure whatsoever, including the misleading endoscopic and barium-contrast studies. As emphasized by Kassirer in the past2 and by the authors, there must come a time when physicians can accept a degree of uncertainty, particularly with regard to attempts to make a definitive diagnosis of a cancer that is often untreatable or incurable.
2 ReferencesJacob Lokich, M.D.
The Cancer Center, Boston, MA 021201
Putterman C ,Ben-Chetrit E . Testing, testing, testing . . .. N Engl J Med 1995;333:1208-1211
Full Text | Web of Science | Medline2
Kassirer JP . Our stubborn quest for diagnostic certainty -- a cause of excessive testing. N Engl J Med 1989;320:1489-1491
Full Text | Web of Science | Medline
To the Editor:
In the Clinical Problem-Solving article “Testing, Testing, Testing . . .,” comfort with diagnostic uncertainty, the risks of treatment weighed against the potential benefits of further testing, and the leading hypothesis being pursued were all suggested as salient variables for the selection of tests. Another variable mentioned was the difference between a hypothetical case and an actual patient. We want to share results from a survey suggesting that this last variable may be a major explanation for the differences between the liberal tester and the parsimonious tester.
One of us saw a 46-year-old man who had a two-year history of burning midepigastric pain relieved by food. Examination demonstrated only a grade 1/6 aortic-insufficiency murmur. The results of laboratory tests were normal except for an alanine aminotransferase level of 89 IU per liter (normal, 0 to 60). Three years earlier the alanine aminotransferase level had been 30 IU per liter. A decision was made to reexamine the patient's liver function in six months on the basis of published recommendations1 and data that 84 percent of results of single tests of liver function showing elevations are falsely positive.2 He was referred to a gastroenterologist for upper gastrointestinal endoscopy, which revealed duodenitis, an active duodenal ulcer, and no evidence of Helicobacter pylori. The gastroenterologist ordered a battery of additional laboratory tests costing $384 to evaluate further the alanine aminotransferase level.
Noting the additional testing, we hypothesized that this represented an example of more expensive specialty care as compared with less expensive primary care and designed a survey to test this. Academic and practicing general internists and gastroenterologists were sent this case report and a list of tests that could be used to evaluate the alanine aminotransferase level further.
To our surprise, specialists and generalists alike indicated they would wait six months, repeat the alanine aminotransferase measurement, or perform only liver-function tests initially. The gastroenterologist who actually treated the patient, when faced with this hypothetical case, ordered only repeated liver-function tests ($61) and indicated that he would delay further testing. This small experience suggests that when faced with a hypothetical case, physicians may find diagnostic uncertainty easier to accept and will carefully consider reasons for ordering tests. In practice, diagnostic uncertainty may be less palatable, and as a consequence, physicians may be more liberal testers. Although we present as evidence the responses of only one person, this interesting dichotomy may help explain some of the differences noted by Putterman and Ben-Chetrit. What we say we would do and what we actually do in clinical problem-solving may differ.
2 ReferencesCynthia G. Kreger, M.D.
Robert A. Murden, M.D.
Ohio State University, Columbus, OH 432101
Kundrotas LW ,Clement DJ . Serum alanine aminotransferase (ALT) elevation in asymptomatic US Air Force basic trainee blood donors. Dig Dis Sci 1993;38:2145-2150
CrossRef | Web of Science | Medline2
Flora KD ,Keefe EB . Significance of mildly elevated liver tests on screening biochemical profiles. J Insur Med 1990;22:206-210
To the Editor:
The insightful discussion by Drs. Putterman and Ben-Chetrit points out the importance of remembering analytic probability before one orders any test. I was dismayed, however, to see the misuse of the serum CA-125 test as a screen for ovarian cancer. The pelvic and rectal examinations were known to be normal, and in a post-menopausal woman, the diagnostic yield for cancer is low for this test.
An additional cause for dismay is the statement that the elevated CA-125 level was assumed to be a false positive result. I believe rather that it was a true positive value for an alternative diagnosis. Several studies, including some of the original ones by Bast and colleagues, identified the presence of this antigen in pleural and peritoneal lining cells.1,2 Several other case reports and clinical studies have documented the increase in CA-125 levels in a variety of benign situations such as ascites, menstruation, and other inflammatory or infiltrative processes involving the mesothelium.2-4 Eosinophilic gastroenteritis should probably be added to the list of diagnostic entities to be considered when the CA-125 is increased.
4 ReferencesLewis A. Hassell, M.D.
Dahl–Chase Pathology Associates, Bangor, ME 044011
Kabawat SE ,Bast RC Jr ,Bhan AK ,Welch WR ,Knapp RC ,Colvin RB . Tissue distribution of a coelomic-epithelium-related antigen recognized by the monoclonal antibody OC125. Int J Gynecol Pathol 1983;2:275-285
CrossRef | Web of Science | Medline2
Apel RL ,Fernandes BJ . Malignant lymphoma presenting with an elevated serum CA-125 level. Arch Pathol Lab Med 1995;119:373-376
Web of Science | Medline3
Kaube A ,Hajiro K ,Adachi Y ,Honda K ,Suzuki T ,Yamamoto T . Lymphangiomyomatosis associated with chylous ascites and high serum CA-125 levels: a case report. Jpn J Med 1987;26:237-242
CrossRef | Medline4
Mastropaolo W ,Fernandez Z ,Miller EL . Pronounced increases in the concentration of an ovarian tumor marker, CA-125, in serum of a healthy subject during menstruation. Clin Chem 1986;32:2110-2111
Web of Science | Medline
To the Editor:
In the Clinical Problem-Solving article by Putterman and Ben-Chetrit, the discussant and the authors question the extent of the workup necessary in cases of very high levels of eosinophilia. Moreover, they all dismiss the diagnosis of eosinophilic leukemia, because all the eosinophils in a complete blood count were mature. Recently, we treated a patient for idiopathic hypereosinophilic syndrome that was later found to be eosinophilic leukemia.
A 39-year-old woman presented with a one-week history of a high fever, cough, myalgia, chest pain, and swelling of the eyelids. She had never traveled outside Europe. Physical examination was normal except for tenderness of the right upper quadrant. A complete blood count was normal except for findings of 71 percent eosinophils, all mature. The absolute eosinophil count was 35,540 per cubic millimeter. The erythrocyte sedimentation rate was 40 mm per hour. Other studies showed only a moderate abnormality of liver function. Stool microscopy detected no ova or parasites. Enzyme-linked immunosorbent assays for fasciola, echinococcus, taenia, toxocara, strongyloides, and trichinella were negative. The serum IgE level was normal. A test for antinuclear antibodies and rheumatoid factor was negative. A chest x-ray film showed patchy pulmonary infiltrates and a questionable left pleural effusion. An abdominal ultrasound examination revealed no abnormalities except for the presence of cholelithiasis. A computed tomographic scan of the sinuses revealed no abnormalities. A bone marrow biopsy showed massive numbers of eosinophils whose development was normal and first-degree fibrosis on the Bauermeister scale. A flow cytometric evaluation of the bone marrow did not detect abnormal populations. A diagnosis of hypereosinophilic syndrome was made, and the patient was treated with amoxicillin, mebendazole, and prednisone. The fever, cough, and myalgia disappeared within days. The eosinophil count dropped to 2370 per cubic millimeter after two weeks, and the pulmonary infiltrates disappeared.
Serologic tests for helminths remained negative. However, symptoms and eosinophilia returned after corticosteroid therapy was stopped. Cytogenetic analysis of the bone marrow unexpectedly showed 30 of 30 cells in mitosis to have an abnormal karyotype: 46,XX,t(9;16)(q34;p13.1), indicating that the correct diagnosis was eosinophilic leukemia.
We conclude that the presence of mature eosinophils does not exclude the possibility of an eosinophilic leukemia and that the diagnosis of the idiopathic hypereosinophilic syndrome should not be made too early.
J. Van den Ende, M.D.
W. Schroyens, M.D., Ph.D.
E. Van den Enden, M.D.
University Hospital, B-2000 Antwerp, BelgiumAuthor/Editor Response
The authors reply:
To the Editor: Van den Ende et al. suggest that our patient may have had eosinophilic leukemia rather than idiopathic hypereosinophilic syndrome. Indeed, as discussed in the excellent review by Weller and Bubley1 cited in our paper, the distinction between eosinophilic leukemia and hypereosinophilic syndrome may be quite difficult to make in some patients. Although cytogenetic analysis was not performed on our patient, chromosomal abnormalities can also occur in hypereosinophilic syndrome.1 The absence of other leukemic features and the clinical course in the months since the diagnosis do not support the hypothesis of a malignant cause of her disease at this time.
Dr. Lokich and Dr. Hassell disagree with the steps taken by the primary physician to exclude a diagnosis of cancer. Dr. Lokich suggests that had cancer been found in this patient, it would probably have been “incurable and untreatable.” We cannot speak for the primary physician (the liberal tester) who chose the specific diagnostic workup; however, we think that confirming a diagnosis of a neoplastic disorder has crucial implications for the patient, even if the disease is not curable with available therapies. The clinical adage that an uncommon manifestation (eosinophilia) of a common disease (cancer) occurs more frequently than a common manifestation of an uncommon disease (hypereosinophilic syndrome) may also have played a part in the formulation of the diagnostic plan. The elevation in the CA-125 level could have been a true positive result for an alternative diagnosis, as pointed out by Dr. Hassell. However, for the purposes of the patient's physicians (and similarly for other physicians who commonly use tumor markers to screen for cancer), the high level of CA-125 was a false positive result that no doubt contributed to a further extension of the diagnostic search.
The preliminary results of the survey conducted by Drs. Kreger and Murden are interesting. Indeed, it would seem to be the experience of many physicians that when they are faced with a critically ill patient in practice, diagnostic uncertainty suddenly becomes much less palatable than when the same patient is discussed in a hypothetical context. We agree with Drs. Kreger and Murden that this dichotomy may explain some of the observed differences between the attending physicians and the discussants in the Clinical Problem-Solving feature.
1 ReferencesChaim Putterman, M.D.
Albert Einstein College of Medicine, Bronx, NY 10461Eldad Ben-Chetrit, M.D.
Hadassah University Hospital, Jerusalem, Israel 911201
Weller PF ,Bubley GJ . The idiopathic hypereosinophilic syndrome. Blood 1994;83:2759-2779
Web of Science | Medline
