v~.t\,t+A-

Improved Detection and Identification of Causes

J. DAVID BESSMAN, M.D.

Department of Internal Medicine and the Pryce-JonesMemorial Laboratory, University of Texas Medical

Branch at Galveston, Galveston, Texas

The peripheral blood film and the red blood cell size distribu-tion histogram were examined for evidence of red blood cellfragments in 2,350 subjects. To distinguish subjects withgreater than or equal to 10 fragments/l,OOO red blood cells(abnormal) from normal, examination of the blood film was0.83 sensitive and 0.30 specific, whereas examination of thehistogram was 0.97 sensitive and 0.87 specific. The most com-mon causes of abnormal fragmentation were malignancy withcytotoxic chemotherapy and severe iron deficiency. In twosubjects, an abnormal red blood cell fragmentation pattern wasthe clue to a spectrin mutant in subjects with an automatedblood count previously evaluated as normal. The data suggesttwo conclusions: the red blood cell volume histogram appearsmore accurate than the peripheral blood film for routine iden-tification of red blood cell fragments; and asymptomatic spec-trin abnormality identifiable by abnormal histogram may be arelatively common disorder. (Key words: Automated bloodcounts; Red blood cell fragments; Spectrin; Chemotherapy)Am J Clin PathoI1988;90:268-273

cells. The cases so described include extreme cases ofburns,4 megaloblastic anemia,s prosthetic heart valve-induced hemolysis,7 DIC,3 or hereditary elliptocytosis orpyropoikilocytosis.'7 It is not known how well the ab-normal histogram pattern correlates with a peripheralblood film judged abnormal for fragments nor how welleither correlates with an actual number of red blood cellfragments. We have compared the examination of theperipheral blood film and of the red blood cell histogramin their accuracy to detect a specific number of redblood cell fragments and have determined the hereditaryand acquired causes of abnormal red blood cell frag-mentation in 93 subjects.ABNORMAL RED BLOOD CELL fragmentation in

the peripheral blood accompanies a large group of dis-orders. In most cases, the fragments reflect an acquiredcause, e.g., cardiovascular abnormality,s.'S.2o microan-giopathic hemolytic anemia,I.13 disseminated intravas-cular coagulation (DIC),21 burns,4 erythroleukemia,3 ormegaloblastic anemia.s In a rare few, the fragments re-flect a hereditary structural abnormality of the red bloodcell membrane protein spectrin.1O.1S The customarymethod to detect red blood cell fragments is by the care-ful examination of the peripheral blood film.24-26 Gener-ally, the evaluation is not quantitative: the impression isrecorded as fragments present or absent. There also havebeen several instances in which evidence of red bloodcell fragments has been described in the histogram of redblood cell size that is now standard in many automatedblood counters. The characteristic pattern is an eleva-tion above baseline of the histogram in the area to theleft (smaller particles) of the peak of intact red blood

Materials and Methods

Peripheral blood films were examined by a trainedtechnologist (minimum experience 15 years) as part ofthe initial blood count for patients at the University ofTexas Medical Branch. The subjects were the inpatientsadmitted to the internal medicine wards of the JohnSealy Hospital and outpatients in the hematology andoncology clinics during July 1985 to March 1986. Onepart of the assessment is whether abnonnal red bloodcell fragments are present; if so, they are noted on thereport. Independently for each specimen, the automatedblood count and differential (Coulter Counter'fl S-PlusIV, Coulter Electronics, Hialeah, FL) were examined bya hematologist. One component of this examination wasthe subjective judgment of whether the red blood cellhistogram indicated an abnormally high number offragments. The histogram analysis assumed that plate-lets appear at the extreme left of the histogram; in nor-mal subjects, there is a return of the histogram to base-line before the increase of the red blood cell peak.6 Redblood cell fragments are described as elevating this in-tennediate zone above baseline.4.s.7

Received November 13, 1987; received--revised manuscript and ac-cepted for publication March 10, 1988.

Supported in part by grants from the National Institutes of Health(CA39182 and CA0770l).

Address reprint requests to Dr. Bessman: Department of InternalMedicine, JSH 4.164, University of Texas Medical Branch at Galves-ton, Galveston, Texas 77550.

RED BLOOD CELL FRAGMENTATIONV~90.No

PeripheralSmearOnly

Patientswith ThisDisease

HistogramOnly Both Total

2192

29144

275191

1839524

II57

15.,

52

33

0000000020

15

25

000 3 3

2372433

17

93

200

02:1:

14

--Malignancy and cytotoxic

chemotherapy.Iron deficiencyBurnsThrombotic thrombocytopenic purpuraAutoimmune hemolytic anemia

Vitamin BI2 deficiencyMalfunctioning heart valveSpectrin mutanttHemoglobin SSHemoglobin S- {3-thalassemiaMyelodysplasiaAcute leukemiaDisseminated intravascular coagulationUnknown

Total

723030

S4

the "silent carrie..' without fragmentation who were not part of the population screened

* The subjects in Figs. 2£ and F.. Tumor types lymphoma, four cases; Hodgkin's disease, three; breast, nine: gastric, three:

lung, three: multiple myeloma, colon, bladder, testicular, one each.t The subjects in Figs ~-D, including the subject with hereditary pyropoikiloc)1osis and

spectrin heterodimer association (performed by T.Coetzer and J. Lawler1o).

For statistical analysis, Student's (-test was used.

return to baseline for at least five channels to the left ofthe main red blood cell distribution curve.

In 171 subjects, two specimens obtained within aweek, and independent of transfusion, were examinedseparately. There was a disparity between the twoblood-film interpretations in eight cases and betweenthe two histogram interpretations in one. For this studythe interpretation of the initial blood count was used.

There were blood specimens from 2,350 patients, forwhich both of the above examinations were done. Onehundred three of these were chosen for quantitativemanual counting of fragments: all those with abnomla1fragments detected by either technique and ten subjectswith no fragments detemlined by either technique. Foreach subject, the number of fragments among 1,000 redblood cells was detemlined by light microscopic exami-nation (I ,OOOX) of a Wright's-stained film of peripheralblood. Fragments were identified as cells like thoseshown by earlier workers2o,26: triangular or helmetshaped and smaller than whole cells. Teardrop- or oval-shaped whole cells were not included. Fragments greaterthan or equal to 10/ I ,000 red blood cells (1.0%) wereconsidered abnomlal.14 Clinical diagnoses were basedon the infomlation obtained independently of this studyexcept in the cases of spectrin abnomlality. Subjectswith abnormal fragmentation who were receivingchemotherapy were not iron, folate, or vitamin B 12 de-ficient. In each case, analysis of red blood cell spectrincontent was by one- and two-dimensional sodium do-decyl sulfate-polyacrylamide gel electrophoresis and

Results

Among 2,350 subjects, 68 (2.9%) were considered tohave abnormal red blood cell fragmentation from analy-sis of the histogram, and 79 (3.4%) were considered tohave abnormal red blood cell fragments from the sub-jective examination of the peripheral blood film. In 39cases, only one of the two subjective evaluations de-tected fragments, whereas in 54 cases both evaluationsdid so. No patient had platelets less than 130 X 109 fL.The most common causes were iron deficiency anemiaand patients with malignancy receiving cytotoxicchemotherapy (Table I). Among the 28 subjects receiv-ing cytotoxic chemotherapy, 25 were alive six monthsafter fragmentation was detected. When measured, fi-brin degradation products were not elevated. None ofthe 28 had apparent DIC develop. The chemotherapygiven was generally multiagent. Five subjects receivedmitomycin; the others received either cytoxan-metho-trexate-fluorouracil or some variation of vincristine,prednisone, alkylating agents, and anthracycline. Thetumor types are listed in Table I; subjects with lym-phoma all had prior staging laparotomy including sple-nectomy.

Among all patients with either iron deficiency anemiaor malignancy with chemotherapy, only a small minor-

270 BESSMAN AJ.C.P.' September 1988

Table 2. Correlation of Red Blood Cell Fragmentation with other Red Blood Cell Variablesin Iron Deficiency and Cytotoxic Chemotherapy

Presence of Increased Fragments

SmearOnly

NoFragments

986:t 3171 :t9

17.6:t 2.3386:t 133

283.6166.76

15.9.20.2394. 266

-99 :73 J

17.3 j359 :'

177116 :t 23-78:t II-

16.5 :t 2.1327:t 125

396:t 3191:t II

19.6 :t 2.4223 :t 96

247122:t 40-91:t 17

15.3:t 3.1-230 :t 96

5111:t 239O:t9

16.8 :t 2.5245:t 121

Iron deficiencynHgb (g/L) (mean :t 2 SD)MCV (11.)RDW (%)Platelets, (X 109 /L)

ChemotherapynHgb (g/L)MCV (11.)RDW (%)Platelets, (XI09/L)

. Different from values in subjects with abnormal fragments detected by both the red blood cell histogram and the peripheral blood smear. P < O.OS.

ity of subjects had increased fragments. Those subjectswith increased fragmentation generally had more signifi-cant anemia than those with the same disorder who didnot have increased fragments (Table 2). Among subjectsreceiving chemotherapy, but not those with iron defi-

.//..A.A....

A..A.A..

..A..A.

.A.A..AA.

AA.A.....A.A."..AA.A...

.A.A..

30

20

A8

..£1A

£1A?8888

..!!:

'i<.)'C~

~

~...

II)-c~ECItV~IL.

66

66.66.66

6666666

6...6666

10 v

'.7

'7V~

'7W

ciency, the red blood cell distribution width (ROW) washigher in subjects with abnormal fragmentation than inthose without.

Other causes were relatively uncommon (Table I).There were 15 cases with fragments noted only on theperipheral smear, in which all the abnormalities inTable I were excluded. In none of those cases was anabnormal histogram described, and in none were theremore than nine fragments/ 1,000 red blood cells (thehighest number seen in normal subjects) in the quanti-tative evaluation of the blood film.

The quantitation of red blood cell fragmentation fromthe peripheral blood film is shown in Figure I. Subjec-tive examination of the peripheral blood film identifiedas abnormal fragmentation all cases with 17 or morefragments per 1,000 red blood cells (1.7%). In contrast,examination of the histogram identified an abnormalpattern in all cases with greater than or equal to 12fragments/ I ,000 red blood cells (1.2%). None of tennormal subjects' blood films was judged to have abnor-mal fragments by any criterion. To identify more than10 fragments/ I ,000 red blood cells, the red blood cellhistogram had a sensitivity of 0.97 and specificity of0.87; the peripheral film was 0.83 sensitive and 0.30specific (Fig. I). At any level of fragments, evaluation ofthe histogram had greater sensitivity and specificity thandid evaluation of the peripheral film.

Four subjects had none of the expected causes of redblood cell fragmentation but had an abnormal histo-gram with a left-side plateau, slight microcytosis, nor-mal hemoglobin, and a peripheral blood smear showingabnormal red blood cell fragments (Figs. 2B, C, E. F;Table 3). All four had normal transferrin saturation,hemoglobin electropheresis, white blood cell and plate-let count, and heart-valve function. All had one to three

. I' -0Blood Film: + + - -Histogram: + - + -

FIG. I. Correspondence of the number of fragments per 1,000 redblood cells with detection of abnormal fragmentation by examinationof the peripheral blood smear and of the red blood cell distributionhistogram; + = detection of fragments; - = no detection; .. = malig-nancy and chemotherapy; . = iron deficiency;. = other causes shownin Table I; 6 = no cause found for fragmentation; 'V' = normal subject.

It 52t 8t 2.0J: 102

21102:t 3187:t 10

21.8 :t 2.9187:t 69

RED BLOOD CELL FRAGMENTATION 271.No )

100 100 100

80 80 80

60 60 60

40 40 40

20 20 20

0 0 0

Relative 30 70. .

110 150 30 70. .

110 150 30.

70 110 150

Frequency, 100 100100

% 80 8080

6060 60

40 4040

20 2020

00 0

70. .

110 150 30 30 70 110 15030. . .

70 110 150

Particle volume, 11FIG. 2. Six subjects with suspected spectrin mutation (see Table 3). The hatched curve shows a red blood cell distribution histogram for a nonnal

subject with an MCV of90 fL and an RDW of 14.5%. Abnonnal fragments are seen as a prominent plateau of particles between the left margin andthe peak of red blood cells in A. The plateau is present but less prominent in B. C. E. and F. No plateau of fragments is present in D

10316113

1311

PyropoikilocytosisAsymptomatic carrierAsymptomatic carrierSilent carrierUnknownUnknown

BBBWWB

FFFMMF

73123131149140126

597676837774

32.115.113.813.414.414.9

Not doneal/6SlIIa/2al/6SNormalNot done

ABCDEF

162665392371

previous automated blood counts with the same abnor-malities but that had been evaluated (including exami-nation of the peripheral blood smear) as normal by ahematologist. When the abnormal red blood cell histo-gram was identified, a red blood cell membrane abnor-mality was suspected, and molecular analysis of the redblood cell membrane proteins was obtained in threecases. Two had a molecular spectrin mutant (Ia/65;lIIa2), and one had no detectable spectrin abnormality.

Two other subjects not part of the 2,350 subjectsscreened for fragmentation were included in the study.One has presumed hereditary pyropoikilocytosis, withchronic significant fragmentation aggravated by 46 °Cincubation, microcytosis and anemia, normal iron satu-

ration and hemoglobin electrophoresis, and bloodsmear with significantly abnormal fragmentation; mo-lecular spectrin analysis was not done (Fig. 2A). Theother is a nominally normal subject whose red bloodcells were used as controls in the study of the othersubjects' spectrin. This subject had a normal automatedblood count and red blood cell histogram but was foundto have a la/65 spectrin mutant (Fig. 2D). These twocases are included to illustrate the automated bloodcount values seen in various molecular abnormalities ofspectrin.

None of the 93 subjects with a histogram and/orblood film that was judged abnormal had an automatedblood count and differential that was otherwise entirely

AJ.CP. September 19te J272 BESSMAN

normal. Seventy-nine had a low hemoglobin, 39 had anabnormal mean cell volume (MCV), and 88 had an ab-normal ROW; 15 had an abnormal white blood cellcount; 23 had an abnormal platelet count, and 19 hadan abnormal mean platelet volume (MPV).

more signs of hemolytic anemia or other systemic signsof now-occult coagulopathy will necessitate long-termobservation.

The mechanism of microangiopathic hemolytic ane.mia in patients with cancer has been extensively consid.ered.2,12.23 Because in the present study there is onlyanemia and fragmentation, it is possible that in such 1

patients these abnormalities simply reflect dyserythro- !

poiesis caused by the cytotoxic chemotherapy. Manycytotoxic drugs cause megaloblastic dyserythropoiesis;folate- or vitamin Bl2-deficiency megaloblastic anemiamay be associated with red blood cell fragmentation,$Dyserythropoiesis may also be the mechanism of frag.mentation in iron deficiency, the second most commoncause in our series. In these subjects the most severeanemia and microcytosis was associated with fragmen-tation, similar to the findings in a recent report. I 1,22 Of

. ---, 'J

lodysplasia.24-26 The relative incidence of these causesdepends on the incidence of these disorders in the testpopulation. Among 25 patients in whom the peripheral

Discussion

The data of the present study indicate the range ofdisorders that cause abnormal red blood cell fragmenta-tion. Two additional conclusions are suggested. First,examination of the red blood cell histogram appears tobe more accurate than examination of the peripheralfilm for the specific task of detecting red blood cell frag-mentation. Second, the frequency of low-grade spectrinmutation may be relatively high.

The most common causes of red blood cell fragmen-tation in the present series-iron deficiency and thecombination of malignancy and cytotoxic chemo-therapy-are not often described in earlier lists ofcauses.24-26 In severe iron deficiency, microcytosis shiftsthe entire red blood cell size distribution to the left(smaller size). Further, thrombocytosis often is present,increasing the histogram channels at the extreme left(Table 2). In such cases the histogram may not return tobaseline, and yet no increase in red blood cell fragmentsis present. However, abnormal fragmentation, mea-sured quantitatively from the peripheral blood film, waspresent in our cases of iron deficiency, with a histogramthat suggested the fragmentation pattern (Fig. I). Thesesubjects all had microcytosis and anemia. In contrast,subjects with early iron deficiency have red blood cellhistograms that are abnormal only in the width of thepeak.4.6.16

A syndrome of microangiopathic hemolytic anemia,with or without uremia, hypoxia, or thrombocytopenia,is well recognized in subjects with adenocarcinoma,especially those with gastric carcinoma or receiving mi-tomycin.2.23 The patients in the current study had nosymptomatic, thrombocytopenic, nor biochemical evi-dence of DIC. Although there was generally moderateanemia, no patient had an elevated reticulocyte countnor required red blood cell transfusion. No particulartumor was characteristically associated with abnormalfragmentation. Rather, the distribution of carcinomasreflected the types seen in an outpatient oncology clinic.Only three subjects with abnormal fragmentation hadgastric carcinoma, and five received mitomycin C.Therefore, red blood cell fragmentation without othersystemic changes did not appear to be strongly asso-ciated with a particular tumor or drug therapy. The in-

~ - -Abnormal spectrin structure has been considered a

rare disorder, with an approximate total of 100 reportedcases in the world. 10 Cases generally are detected begin-

ning with the even rarer but clinically more dramatichereditary pyropoikilocytosis. Relatives of the index pa-tients are then studied, revealing the "asymptomaticcarriers" with a lesser molecular and phenotypic abnor-mality.10 By detection of an abnormal red blood cellhistogram, we detected two new cases of a spectrin mu-tant in 2,350 samples. This is too small a number ofcases to allow us to make statistical inferences. However,the random detection of an asymptomatic subject with aspectrin mutant without fragmentation further suggeststhat spectrin mutants may be relatively common in pop-

- . . To

l\ctermine whether any of these patients will manifest

273RED BLOOD CELL FRAGMENTATION.No 3

, other than the one studied here. True "silent;. have blood counts indistinguishable from nor-

~-I (Fig. 2D), whereas "asymptomatic carriers" are:::ntifiable by microcytosis and an abnormal red blood~I histogram (Figs. 2B and C). Not all subjects withlitis pattern have a spectrin abnormality detectable bycurrent techniques (Fig. 2£). However, in the absence ofother more obvious causes of red blood cell fragmenta-tion (Table 1), slight microcytosis and an abnormal redblood cell histogram should prompt suspicion of a mu-

tAnt spectrin.Comparison of the blood film and the histogram eval-

uation suggests that the histogram is both more sensitiveand more specific in detecting abnormal degrees of redblood cell fragmentation. In a less-selected populationwith a low frequency of red blood cell fragmentation,the specificity rate would probably be higher for eithertechnique. From the data in this study, it appears thatthe superiority of the histogram analysis over blood filminterpretation would remain in a general population.Statistically, more precision would be expected of aIO,OOO-cell analysis (histogram) then of a I,OOO-cellquantitative examination or an indeterminate qualita-

tive analysis.Evaluation of the red blood cell histogram for frag-

mentation does not replace the examination of the pe-ripheral blood smear for other purposes. Also, becausean abnormal histogram or smear was associated with anotherwise abnormal automated blood count, use of his-togram analysis would not affect the strategy of per-fotnling visual white blood cell differentials only whenthe automated blood count is abnormal. 19 However, for

case finding of unsuspected red blood cell fragmenta-tion, analysis of the histogram appears superior to rou-tine examination of the peripheral blood film. Thecauses (besides spectrin mutants) such unsuspectedcases will include remain to be determined in largerstudies.

Acknowledgments. The spectrin analysis and helpful interpretationof the abnormalities detected were generously provided by Drs. Ther-esa Coetzer. Jack Lawler, and Jiri Palek. The routine blood smearinterpretation was done by Doris Najar, MT (ASCP), Shirley Bridges,MT (ASCP), and Anne Terrebonne, MT (ASCP). The manuscript was

typed by Mary McChesney.

2. Antman KH, Skarin AT, Mayer RJ, et al: Microangiopathic he-molytic anemia and cancer: a review. Medicine 1979;58:377-384.

3. Atkins JM, Muss HB: Schistocytes in erythroleukemia. Am J MedSci 1985;289:110-113.

4. Bessman JD: Microcytosis caused by red cell fragmentation.JAMA 1977;238:2391-2392.

5. Bessman JD: Erythropoiesis during recovery from macrocyticanemia. Blood 1977;50:995-1000.

6. Bessman JD: Automated blood counts and differentials. Balti-more. Johns Hopkins University, 1986.

7. Bessman JD, Johnson RK: Erythrocyte size distribution in normaland abnormal subjects. Blood 1976;46:336-345.

8. Brodeur MTH, Sutherland DW, Koler RD, Griswold HE: Hemo-lytic anemia and valvular heart disease. N Engl J Med

1965;22:104-105.9. Cantrell JE, Phillips TM, Schein PO: Carcinoma-associated he-

molytic-uremic syndrome. A complication of mitomycin withchemotherapy. J Clin OncoI1985;3:723-734.

10. Coetzer T, Lawler J, Prchal JT et al: Molecular determinants ofclinical expression of hereditary elliptocytosis and pyropoikilo-cytosis. Blood 1987;70:766-772.

II. Cornbleet PJ, Kessinger S: Accuracy of low platelet counts on theCoulter S-Plus IV. Am J Clin PathoI1985;83:78-80.

12. Doll DC, Weiss RB: Neoplasia and the erythron. Jain Onco!1985;3:429-446.

13. Lieberman E, Heuser E, Donnell GN, et al: Hemolytic uremicsyndrome. NEngl J Med 1966;275:227-236.

14. Lohrmann HP, Adam W, Heymen B, Kubanek B: Microangio-pathic hemolytic anemia in metastatic carcinoma. Ann InternMed 1973;79:368-375.

15. Marsh GW, Lewis SM: Cardiac haemolytic anemias. Semin He-

matoI1969;6:133-149.16. McClure S, Custer EM, Bessman JD: Improved detection of early

iron deficiency. JAMA 1985;253:1021-1023.17. Mentzer WC, Turetsky T, Mohandas N, et al: Identification of the

hereditary pyropoikilocytosis carrier state. Blood 1984;63:1439-1446.

18. Palek J, Lux S: Red cell membrane skeletal defects in hereditaryand acquired hemolytic anemias. Semin Hematol1983;20: 189-224.

19. Pierre R V: The routine differential leukocyte count vs automateddifferential counts. Blood Cells 1985;11: 11-20.

20. Ravenel SD, Johnson JD, Sigler AT: Intravascular haemolysisassociated with coarctation of the aorta. J Pediatr 1969;75:67-74.

21. Rosner F, Rubenberg ML: Erythrocyte fragmentation in con-sumption coagulopathy. N Engl J Med 1969;280:219-220.

22. Savage RA, Hoffman GC: Spuriously high platelet counts (letter).Am J Clin Pathol 1985;84:406-407.

23. Sheldon R, Slaughter R: A syndrome of microangiopathic hemo-lytic anemia, renal impairment, and pulmonary edema inchemotherapy-treated patients with adenocarcinoma. Cancer

1986;58:1428-1436.24. Spivak JL: Normal hematopoiesis. In: Harvey AM, et ai, eds. The

principles and practice of medicine. 21 st ed. Norwalk: Apple-ton-Century-Crofts, 1984.

25. Williams WJ et al: Hematology. 3rd ed. New York. McGraw-Hill,1983.

26. Wintrobe MM: Clinical hematology. 6th ed. Philadelphia. Lea &Febiger, 1981.

ReferencesAmorosi EL, Ultmann JE: Thrombotic thrombocytopenic pur-

pura. Medicine 1968;45:139-159.