ANNALS OF CLINICAL AND LABORATORY SCIENCE, Vol. 22, No. 2 Copyright © 1992, Institute for Clinical Science, Inc.

Reticulocyte Counting by Flow Cytometry A Comparison with Manual Methods* ALEX A. PAPPAS, M .D ./fi REBECCA B. OW ENS, B.S.,t and JAMES T. FLICK , M .D .t D e p a rtm e n t o f P a th o lo g y f a n d C lin ic a l L a b o ra to rie st, U n ive rsity o f A rka n sa s f o r M ed ica l S cien ces, L ittle R ock, A R 72205

ABSTRACT T he reticulocyte count (RC) is a key diagnostic test in the evaluation, classification, and response to therapy of anemia. T he RC, as determ ined by m anual m ethods, has a frustrating inherent im precision owing to its binom ial counting statistics (i.e., low counts/low precision) and inaccu­ racy because of inter- and intraobserver variability as to w hat in d ee d is a reticulocyte. F luorescent activated cytom etric (FACS) analysis of reticulocytes by thiazole orange (TO) is a rapid, relatively sim ple, and precise m ethod for counting reticulocytes. T he autom ated m ethod counts 10,000 cells or more vs. 1,000 cells counted by the m anual m ethod. Although inherently more precise, the FACS m ethod may be inaccurate owing to the presence of H ow ell-Jolly bodies, nucleated red blood cells (RBCs), sickled cells, or giant platelets. T he RC by FACS is w ell correlated with the m anual m ethod and the reference ranges are similar. A new param eter by FACS, the reticulocyte m aturation index (RMI), provides an in d e p e n d e n t m ea su rem e n t o f reticulocyte RNA content. Although the RMI does not correlate w ith RC either by FACS or m anual m ethods, it does provide an in d ep e n d e n t param eter of erythropoietic activity and may be useful in predicting bone marrow engraftm ent or further subclassifying anemias. D eterm ination by FACS of the RC offers significant advantages over m anual m ethods in m onitoring a patients erythropoietic response. H ow ­ ever, one m ust be cognizant of potential pitfalls in the m ethod.

th e re is gradual c o n d e n sa tio n o f th e n u c le a r c h ro m a tin w ith c o n c o m ita n t D uring the normal m aturation of eryreduction of both the nuclear and total throid precursors in the bone marrow, cell size. W hen th e n u cleu s becom es pyknotic, it is extruded. Sim ultaneously, * Send reprint requests to Alex A. Pappas, M.D., there is cytoplasmic synthesis of hem o­ Director of Clinical Laboratories, University of Arkansas for Medical Sciences, 4301 W. Markham globin. T he hem e portion of hem oglobin St., Mail Slot 502, Little Rock, AR 72205. is synthesized in the m itochondria, and 125 Normal Erythropoiesis

0091-7370/92/0300-0125 $01.20 © Institute for Clinical Science, Inc.

126

PAPPAS, OWENS, AND FLICK

the globin chains separately on the poly­ ribosom es. C ovalent linkage of these two m oieties then occurs in the m itochondria. T he RNA containing polyribosom es can rem ain in the now anucleated erythro­ cyte for up to four days. O ver this four day period of tim e, there is progressive dim i­ nution in the num ber of polyribosom es a n d h e n c e h e m o g lo b in s y n th e s i s . Approximately 25 percent of hem oglobin is synthesized in the reticulocyte or the equ iv alen tly polychrom atophilic annucleate stage of erythroid developm ent. T h ese early an u cleated polyribosom erich erythrocytes have a greater cell vol­ um e than m ature erythrocytes and thus a p p e a r m acrocytic. T h e se m acrocytic e ry th ro cy tes w ill stain polychrom atophilically w ith standard W rights-Giem sa stains since hem oglobin synthesis has not b e e n com pleted.12 Prim ary reticulocyte m aturation (three days) occurs in the bone m arrow. The less th an com pletely m ature erythroid cell th en traverses the m arrow sinusoid and norm ally com pletes its final m atura­ tion (one day) in the peripheral blood. W ith the normal progressive loss of the p ro te in s y n th e s iz in g p o ly rib o so m e s, there is a cessation of hem oglobin syn­ thesis, and transform ation from reticulo­ cyte to norm al m ature RBC. In normal peripheral blood stained w ith W rightsGiemsa, the m ore m ature late reticulo­ cytes are virtually indistinguishable from totally m ature erythrocytes. H ow ever, th e a p p e a ra n c e o f e a sily id e n tifia b le “ p o ly c h ro m a to p h ilic m a c ro c y te s ” or “ shift cells” by W rights-G iem sa staining, generally reflects the prem ature release o f reticulocytes from th e m arrow. T he appearance of such “ shift” reticulocytes indicates accelerated erythropoiesis and carries significant diagnostic importance. M orphologic D efinition of th e Reticulocyte W hen the polyribosom e-rich red cells have ju st lost their nucleus, they can be

stained w ith supravital stains such as New M ethylene Blue (NMB) or B rilliant Cresyl Blue (BCB). If two or m ore parti­ cles of basophilic polyribosom al m aterial are present, the cell is term ed a reticulo­ cyte.8 Normal young reticulocytes may contain m uch polyribosom al m aterial, w hich is progressively lost as they con­ tinue to m ature to norm al red cells. A m orphologic classification schem e for such m aturation has b e e n developed, b a sed on H e ilm e y e r’s orig in al w ork, revealing five stages.9 This schem e is now the generally accepted m orphologic standard for m anual reticulocyte count m easurem ent. A biochem ical rather than a m orphological definition of a reticulo­ cyte could be that of an anucleated eryth­ rocyte containing seven to 17 p e rc e n t o f th e o rig in a l e ry th r o b la s t r ib o n u ­ cleic acid.10 M anual Reticulocyte Count M easurem ent T he standard 1,000 cell m anual reticu­ locyte count has a notoriously high coef­ ficient of variation (CV) because o f its in h e re n t im p re cisio n an d inaccuracy. C o n trib u tin g factors in clu d e ; in te ro b ­ server variation in th e d efinition o f a reticulocyte, staining variability, variable distribution of cells on the slide, and sta­ tistical sam pling error.13 T he reticulocyte count statistics follow a binom ial distri­ bution, w hich results in unaccep tab le precision for reticulocyte counts less than approxim ately one percent, i.e., w ithin the normal range. F urther, this range of m easurem ent is w here detection of early erythroid activity w ould be m ade, such as ju st after a bone m arrow transplant, ju st after chem otherapy, correction of iron, B12 or folate deficiency, or in a sickle patient recovering from an erythroid cri­ sis. This statistical deficiency can be cor­ rected by counting m ore cells (as noted in table I).11 H ow ever, the counting of the required 10,000 red cells by m anual

127

RETICULOCYTE COUNTING WITH FACS

TABLE I Flow Cytometric Versus Manual Reticulocyte Count 95 Percent Confidence Limits Expected Reticulocyte Counts and CV's Actual ; Retie %

Flow Cytometric [10,000 cells] [30,000 cells]

Manual [1000 cells] Range 0-3

CV 100

1

0.4-1.6

32

0.8-1.2

5

3.6-6.4

14

4.6-5.4

10 15

0.1

Range CV 0.04-0.16 22

Range 0.07-0.14

CV 18

10

0.9-1.1

5.7

4.2

4.8-5.2

2.8

8.2-11.9 9.5

9.4-10.6 3.0

9.6-10.3

1.7

12.8-17.4 7.5

14.3-15.7 2.4

14.6-15.4

1.4

*Normal range 0.5 - 1.5 % (Adults)

Statistical comparison of a single retie count measurement, either by a manual or flow cytometric method. The flow cytometric method permits useful measurement below a retie count of 1%, making it a useful indicator of early erythrold recovery, e.g.. post bone marrow transplant. For a description of the binomial distribution model used in these calculations, see: Greenberg, E.R. and Beck, J.R. : Arch. Pathol. Lab. Med. 108:396-398. 1984. m ethods w ould require a nearly superhu­ m an effort. O th e r in accu racies in the m anual m ethod include potential errors from m iscounting H ow ell-Jolly bodies (deoxyribonucleic acid rem nants), H einz bodies (oxidized or denatured hem oglo­ bin), P a p p e n h e im e r b o d ies (siderotic granules), or H em oglobin H inclusions as precipitated reticulum . Clinical Interpretation of the Reticulocyte C ount An increased peripheral blood reticu­ locyte count is a sensitive and reliable in d ic a to r o f in c re a s e d h e m a to p o ie tic activity and is an essential elem en t in the diagnostic workup o f the various ane­ mias (figure 1). F o r clinical purposes, re tic u lo c y te s are com m only re p o rte d

as a p e rc e n ta g e of th e e ry th ro c y te s. F o r a d u lts , th e r e tic u lo c y te c o u n t is commonly given as 0.5 p ercen t to 1.5 percent w ith a slightly higher range for wom en, and 2.5 percent to 6.5 percent for newborns. An in h eren t feature of the reticulocyte count percentage is that it is m ade rela­ tive to the m ature red cell population and, consequently, may appear increased w hen there is a significant reduction in the num ber of m ature erythrocytes. Inex­ perienced physicians may not know that an apparently elevated reticulocyte (per­ centage) count may not be elevated w hen correction is m ade for a low red cell count owing to severe anem ia. This problem is fairly com m on, as th e p a tie n ts m ost closely exam ined for erythroid recovery are often severely anemic. In such cases,

128

PAPPAS, OWENS, AND FLICK

FIGURE 1.

A n a lg o r ith m u tiliz in g t h e r e tic u lo c y te c o u n t in t h e e v a lu a tio n o f a n e m ia .

or in periods of intense erythropoiesis, the percentage reticulocyte count is inac­ curate and a “correction” can be made (table II).15 First, an absolute reticulo­ cyte count can be calculated by m ultiply­

ing the percentage reticulocyte count by the RBC count. Secondly, the reticulo­ cyte count may b e “ corrected” for the d e g re e of a n e m ia by c o m p a rin g th e patient’s hem atocrit to a normal hem ato­

TABLE I Various Corrections Used in Reporting Reticulocytes

Method

Formula

Absolute Count

Reticulocytes (%) x RBC Count

Reticulocyte Index

Reticulocvtes (%) x Patient's Hct 0.45

Reticulocyte Production Index

Reticulocvtes (%) x Patient's Hct Maturation Time (days) 0.45 1.0 days when Hct = 45 1.5 days when Hct = 35 2.0 days when Hct = 25 2.5 days when Hct = 20

The various reticulocyte 'corrections' can be used to categorize anemia on a kinetic basis.

RETICULOCYTE COUNTING WITH FACS

c rit: n a m e ly th e r e tic u lo c y te in d ex . Finally, since the percentage o f reticulo­ cytes can be altered by the presence of “ shift” reticulocytes in d icatin g prem a­ ture release from the marrow, the reticu­ locyte production index (RPI) has been proposed to correct for both the “ shift” and anemia. Flow Cytom etric R eticulocyte C ount M easurem ents A u to m a te d m e th o d s fo r c o u n tin g reticulocytes, w hich can overcom e many of the previously-m entioned lim itations of the m anual m ethods, include either image analysis of blood films using auto­ m ated m icroscopy or flu o re sc e n t acti­ vated flow cytom etric analysis (FACS) m ethods.7 In the FACS m ethods, a fluo­ re s c e n t dye, such as th ia z o le orange (TO), acridine orange (AO), or pyronin Y, com bines w ith ribosom al RNA in reticu­ locytes. T he degree of fluorescent em is­ sion c a u se d by th e ex c itatio n o f th e bound dye is proportional to the am ount of RNA in an e ry th ro cy te and h en ce inversely to its maturity. Because it is a sim ple staining proce­ dure and good resolution is obtained, TO is becom ing the fluorescent dye of choice for reticulocyte counting in flow cytom e­ try.3 T he acquisition and analysis of the stained samples by FACS are relatively sim ple. By settin g th e forw ard-scatter (FSC) and side-scatter (SSC) am plifier gains to th e “ L O G ” m ode, th e w hite blood cell (WBC) population, w hich will also stain with the TO, is “off scale” and therefore not included in the data acqui­ sition. T he RBC/reticulocyte population will be the major population “on scale.” After acquisition, a gate is draw n around the RBC/retic population w hich further excludes any small WBCs, platelets, and debris seen. This gated population can th en b e assessed for p e rc e n t positive

129

reticulocytes (FL-1, green fluorescence) based on a paired control sam ple for each patient (figure 2—A, C, E). T h e re are now so ftw a re p ro g ra m s available that further simplify the statis­ tical analysis used in FACS reticulocyte counting.2 After appropriate instrum ent c a lib ra tio n , such as w ith C aliB R IT E B eads,1 the reticulocyte softw are uses those calibrations as a starting point and autom atically sets up th e a p p ro p riate instrum ent settings and gains. U sing the F S C v s. S S C d o t p l o t, t h e R B C / reticulocyte population is adjusted to the upper right hand corner of the dot plot and a FSC threshold is set. After acquisi­ tion of 10,000 to 50,000 events (counts), the RBC/reticulocyte gate is established. This gate should be greater than 85 per­ cent of the total population in the dot plot. U sing an in tern al algorithm , the software establishes a break point on the right side of the baseline of th e m ain negative cell population for each patient. From this point the positive cell popula­ tion can be determ ined, w hich results in a histogram display sim ilar to figure 2-E. This autom atic calculation elim inates the n eed for a negative control tube to be run w ith each patient. A FL-1 green fluores­ cence histogram , w ith a m arker draw n and the percent positive and the m ean f lu o r e s c e n t c h a n n e l s ta te d , is th e n printed. If the RBC count is provided, an absolute reticulocyte count can autom ati­ cally be obtained. T he autom ated FACS m ethod elim i­ nates m uch of the statistical variability in reticulocyte counting; how ever, som e erythrocyte elem ents, such as H ow ellJolly bodies or nucleated erythrocytes, may cause false positive results. These cells usually do not interfere if the w in­ dow gates and fluorescence thresholds are properly set. Some nonerythroid con­ ditions, such as chronic lymphocytic leu­ kemia or conditions w ith giant platelets, may also cause spurious resu lts.14 T he

130

PAPPAS, OWENS, AND FLICK

Tèa

té» "Té*

B DÉBRIS

’s-

-té«.....lé* F S C --->

,:

"ïé» '

"ft1.....id*'

FLl^R*t-COUNT--->

F ig u r e 2. Dot Plot A shows a normal forward-scatter count (FSC) vs side-scatter count (SSC) pattern of the red blood cell/retic population which results in a normal FL1 vs FL2 dot plot (C) and a normal FL-1 histogram (E) when the appropriate gate is activated. Dot Plot B, in contrast, shows an abnormal scatter pattern due to increased debris. This can cause suboptimal gating. Dot Plot D is an example of the FL1-FL2 green and red fluorescence which shows some interfering components that can result in an abnormal histogram (F ).

operator m ust be alert to abnorm al histo­ gram distributions and gate appropriately to avoid p o te n tia lly erroneous results from patients w ith these conditions (fig­ ure 2-B , D, F). A review of a W rights-Giem sa stained slide and/or a m anual reticulocyte sm ear evaluation should be perform ed on all su sp icio u s FACS histogram s. O nce a patient has b e e n determ ined to give dis­ c re p a n t re tic u lo c y te re s u lts by flow cytom etry, w hich cannot be corrected by o perational gate changes, th en only a m anual reticulocyte count should be per­ formed. T he incidence of patients w ith d iscrep an t results b e tw e en FACS and m anual reticulocyte counting m ethods will d e p e n d on the patient population of each respective institution. S in ce th e flow c y to m e tric m e th o d counts 10,000 or m ore cells (now 30,000 cells in our laboratory) in contrast to a

1,000 cell m anual reticulocyte count, the im p ro v e d sta tistic a l p re c is io n of th e FACS m ethod is w ell docum ented (table I).14 T he accuracy of the FACS m ethod for correctly id en tify in g reticulocytes, how ever, may be questioned relative to m anual m ethods. In the m anual NMB m ethods, two particles of blue staining reticulum m ust be seen before a cell is counted as a reticulocyte, w hereas the sensitivity of the FACS m ethod is such th a t c e lls w ith m u ch less re tic u lu m w ould be detected. It is also possible that the am ount of reticulum precipitated by NMB m ay n o t b e fully e q u iv a le n t to the RNA bound by the various fluores­ cent dyes. T he Reticulocyte M aturation Index T h e flu o re s c e n c e in te n s ity o f th e TO -stained reticulocyte is proportional to

131

RETICULOCYTE COUNTING WITH FACS

TABLE III Comparison of CAP Workload Recording of Manual Versus Fluorescent Activated Cytometric Reticulocyte Counting

Method MANUAL Mix with NMB = 5-10 min. Prepare Smear Dry = 5-10 min. Count - 1,000 CELLS Report FLOW Calibrate Instrument = 7 min. Load Program & Patient List = 3 min.

CAP Units

CAP Technical Time

9

9 min/1 sample 90 min/10 samples

11

11 min/1 sample 22 min/10 samples

5ul Blood + 1ml TO = 30 sec. Incubate 30 min. Acquire and analyze 1 sample = 45 sec. 10 samples = 4.5 min.

Comparison ot time and effort of manual versus automated reticulocyte counting. The automated method clearly reduces technical time for batch analysis of reticulocytes and is not advantageous for single analysis. the am ount of RNA w ithin the cell, w hich in turn is a function of the m aturity of that cell. M easurem ent of this fluorescence intensity has b e e n term ed as the reticu­ locyte m aturation index (RM I).4,5 T he RM I is d eriv ed from th e fluorescence histogram by quantifying the geom etric m ean fluorescence intensity of the posi­ tiv e c e lls . T h e r e is n o c o r r e la tio n b etw een the RM I and reticulocyte p er­ centage by FACS in healthy adults or hospitalized patients. T he RM I appears to be an in d ep en d en t and sensitive indi­ cator of erythroid activity.4,5 T he RM I m easu rem en t provides an in d ep e n d e n t objective m easurem ent of b o n e m arro w e n g ra ftm e n t fo llo w in g transplantation. Engraftm ent is currently determ ined principally by the rise in the

absolute n e u tro p h il count (ANC). T he ANC and RMI rise in parallel in early uncom plicated bone m arrow transplant recovery. Although the ANC is a sensi­ tive indicator of bone m arrow engraft­ m ent, the ANC can drop during episodes of clinical or subclinical infection and could b e m isleading as to the success of transplantation. In contrast, the RM I is essentially in d ep en d en t of an infectious process. T he RMI can also b e used to su b c la ssify an e m ias, p a rtic u la rly th e h y p o p ro life rativ e an em ias w h e re th e reticulocyte count is low, or as an indica­ tor of erythropoietic activity follow ing therapy for anem ia.5 T he reticulocyte count by FACS does not lend itself to random individual sam­ ples since the procedure is m ore labor

132

PAPPAS, OWENS, AND FLICK

in te n s iv e w ith re s p e c t to in stru m e n t set-up tim e and is b est suited for batch analysis (table III). T he instrum entation is relatively expensive at this tim e and is not suitable for low volum e laboratories. C onclusions I n s u m m a ry , th e flo w c y to m e tric approach to th e reticulocyte (percentage) count greatly expands the utility of the test by expanding the useful operational range of th e test to w ell below the normal range. T h e lim itatio n s of th e m anual m eth o d in m e a su rin g norm al or low reticulocyte counts are essentially elim i­ nated, perm itting detection of early erythroid bone marrow recovery in oncology patients, and increasing the sensitivity of the detection of early erythroid recovery in a variety of other hem atologic disor­ ders. T h e RMI appears to be a separate, clinically useful indicator of bone m ar­ row recovery following transplantation or chem otherapy and in the evaluation of erythroid activity in anem ic conditions. R eferences 1. Becton Dickenson Immunocytometry Systems: FACScan " AutoCOMPSoftware. 2. Becton Dickenson Immunocytometry Systems: Retic-Count " Reticulocyte Enumeration Soft­ ware. 3 . C a r t e r , J. M ., M c S w e e n e y , P. A ., W a k e m , P. J., and NEM ET, A. M .: Counting of reticulo­ cytes by flow cytometry: Use of thiazole orange. Clin. Lab. Haemat. 11:567-571, 1989. 4. DAVIS, B . H. and BIGELOW, N. C.: Flow cyto­ metric reticulocyte quantification using thia-

5.

6.

7.

8.

9. 10.

11.

12. 13.

14.

15.

zole orange provides clinically useful reticulo­ cyte maturity index. Arch. Pathol. Lab. Med. 113:684-689, 1989. D a v is , B. H., B i g e l o w , N. C ., B a l l , E. D ., M i l l s , L ., and C o r n w e l l , G . G .: Utility of flow cytometric reticulocyte quantification as a predictor of engraftment in autologous bone marrow transplantation. Am. J. Hemat. 32:81— 87, 1989. D e i s s , A. and K u r t h , D .: Circulating reticulo­ cytes in normal adults as determined by the new methylene blue method. Am. J. Clin. Pathol. 53:481-184, 1970. F e r g u s o n , D. J., L e e , S. F ., and G o r d o n , P. A.: Evaluation of reticulocyte counts by flow cytometry in a routine laboratory. Am. J. Hemat. 33:13-17, 1990. F i n c h , C. A., M a r s h a l l , P. N., B r e c h e r , G ., T r o b a u GH, F . E., and ASSEND ELFT, O. W .: Method for reticulocyte counting. National Committee for Laboratory Standards. 10:225235, 1985. G i l m e r , P. R. and K o e p k e , J. A.: The reticulo­ cyte. Am. J. Clin. Pathol. 66:262-267, 1976. G r a s s o , J. A., W o o d a r d , J. W ., and S w i f t , H.: Cytochemical studies of nucleic acids and pro­ teins in erythrocytic development. Zoology 50:134-140, 1963. G r e e n b e r g , E. R. and B e c k , J. R.: The effects of sample size on reticulocyte counting and stool examination—The binomial and poisson distributions in laboratory m edicine. Arch. Pathol. Lab. Med. 108:396-398, 1984. K o e p k e , J. F. and K o e p k e , J. A.: Reticulocytes: A Review. Clin. Lab. Haemat. 8:169-179,1986. S a v a g e , R . A., S k o o g , D. P., and R a b i n o VITCH, A.: Analytic inaccuracy and imprecision in reticulocyte counting: A preliminary report from the College of American Pathologists reticulocyte project. Blood Cells il:9 7 -1 1 2 , 1985. V a n H o v e , L., G o o s s e n s , W., V a n D u p p e n , V ., and VERWILGHEN, R. L.: Reticulocyte count using thiazole orange: A flow cytometry method. Clin. Lab. Haemat. 12:287-299, 1990. W i n t r o b e , M. M., L e e , G . R., and B o o g s , D. R.: Hem olytic Disorders. In: C linical Hematology, 8th ed. Philadelphia, Lea and Febiger, 1981, pp. 744—745.