Proc. Natl. Acad. Sci. USAVol. 89, pp. 2694-2698, April 1992Medical Sciences

Reverse transcription polymerase chain reaction for the rearrangedretinoic acid receptor a clarifies diagnosis and detects minimalresidual disease in acute promyelocytic leukemiaWILSON H. MILLER, JR.*, AKIRA KAKIZUKAt, STANLEY R. FRANKEL*, RAYMOND P. WARRELL, JR.t,ANTHONY DEBLASIO*, KRISTI LEVINE*, RONALD M. EVANSt§, AND ETHAN DMITROVSKY*¶*Laboratory of Molecular Medicine, and tLeukemia and Developmental Chemotherapy Services, Department of Medicine, Memorial Sloan-Kettering CancerCenter, New York, NY 10021; and tGene Expression Laboratory, The Salk Institute for Biological Studies and §Howard Hughes Medical Institute, San Diego,CA 92186

Contributed by Ronald M. Evans, December 31, 1991

ABSTRACT The characteristic t(15;17) of acute promye-locytic leukemia (APL) fuses the retinoic acid receptor a(RAR-a) gene on chromosome 17 to a gene on chromosome 15called PML, a putative transcription factor. This distincttranslocation results in a fusion mRNA detected by Northernanalysis. Two cDNAs have been isolated that differ in the extentof 3' PML nucleic acid sequence contained. This study de-scribes a reverse transcription polymerase chain reaction (RT-PCR) assay for the PML/RAR-a fusion transcript, whichamplifies PML/RAR-a mRNA from APL cells with eitherreported breakpoint. DNA sequencing of the predominantRT-PCR products from 6 patients showed identical RAR-aexonic breakpoints and two PML breakpoints. This RT-PCRassay was positive in leukemic cells from 30/30 APL patientswith the molecular rearrangement confirmed by cytogeneticsor Northern analysis. In leukemic cells of patients with amorphologic diagnosis of APL lacking the t(15;17) by routinecytogenetics, a positive RT-PCR assay predicted clinical re-sponse to all-trans-retinoic acid (RA) therapy. Dilutional stud-ies with leukemic cells that express (NB4) or do not express(HL-60) a PML/RAR-a fusion mRNA reveal that this RT-PCR assay detects the transcript from as little as 50 pg of totalRNA. In APL cells from 5/6 patients treated with RA alone,a complete response by clinical and cytogenetic criteria accom-panied a persistently positive RT-PCR assay. This precededrelapse by 1-6 months. RT-PCR for PML/RAR-a mRNAprovides a more-sensitive test for the t(15;17) than routinecytogenetics or Northern analysis. This molecular rearrange-ment detected by RT-PCR best defines this RA-responsivemalinancy. The RT-PCR assay for the PML/RAR-a tran-script yields important diagnostic and prognostic informationin the management of APL patients.

Acute promyelocytic leukemia (APL) is a distinct clinical andhistopathologic subtype of acute myeloblastic leukemia.APL is associated with a unique cytogenetic abnormality,t(15;17)(q22;ql2-21) (1). Recognition of this subtype of leu-kemia is important because of its potential for life-threateningdisseminated intravascular coagulation (2) and the dramaticclinical response to all-trans-retinoic acid (RA) treatmentlimited to this form of leukemia (3-5).The characteristic t(15;17) ofAPL creates a fusion between

a newly described gene (PML) on chromosome 15 and the RAnuclear receptor a (RAR-a) (6-8). The fusion gene is ex-pressed at the mRNA level in leukemic cells isolated fromAPL patients (5, 9-11). The cDNA clones from several APLcellular samples have been isolated and sequenced, revealingat least two different exonic fusion sequences (7, 8, 12). The

breakpQint in the RAR-a gene ofAPL cells is found uniformlybetween the first and second coding exons. Two fragments ofthe PML gene, differing in their 3' region, contribute to therearranged PML/RAR-a cDNA (7, 8, 12).These molecular findings have been used to examine

leukemic cells of patients for the presence of the PML/RAR-a rearrangement. Northern analysis can detect abnor-mal RAR-a in most clinical APL cases if a sufficient quantityof intact RNA is obtained from the leukemic cells (5, 9-11).When adequate genomic DNA is obtained, Southern analysiswith exonic probes detects a fraction ofAPL cases (5, 10, 11).The use of multiple intronic probes for both RAR-a and PMLallows detection of most, if not all, rearrangements (13, 14).Reverse transcription polymerase chain reaction (RT-PCR)with primers derived from the cloned PML/RAR-a se-quences has amplified the fusion molecule in selected cases(7, 8, 12).

This study describes a RT-PCR method that detects bothreported PML/RAR-a rearrangements. It applies this mo-lecular test to leukemic cells from 36 patients with a clinicaldiagnosis ofAPL and compares its sensitivity and specificitywith that of cytogenetics and Northern analysis. The findingsreveal that the detection ofthe PML/RAR-a by RT-PCR bestpredicts the clinical response to RA therapy. This assay hassufficient sensitivity to detect minimal residual disease inpatients who are clinically in complete remission (CR) afterRA treatment.

MATERIALS AND METHODSNorthern Analysis of Leukemic Cells. Total RNA from

leukemic cells of APL patients was prepared as follows:mononuclear cells were separated from bone marrow aspi-rates by Ficoll/Hypaque density centrifugation. Total cellu-lar RNA was prepared by ultracentrifugation in a 4 Mguanidine isothiocyanate/5.7 M cesium chloride gradient byestablished techniques (15). This technique provides a highyield of intact RNA from these RNase-enriched cells (5, 9).Northern analysis was performed with transfer to reinforcednitrocellulose filters by standard techniques (16). The ob-tained filters were hybridized to [a-32P]dCTP radiolabeledDNA probes and washed as described (9). Autoradiographywas with Kodak XAR film and exposure to an intensifyingscreen at -70°C. The RAR-a probes were either a 600-base-pair (bp) Pst I or a 640-bp EcoRI/Sst I-cut insert purified froma plasmid containing the full-length RAR-a cDNA (17, 18).

Abbreviations: APL, acute promyelocytic leukemia; RA, all-trans-retinoic acid; RAR-a, RA receptor a; RT-PCR, reverse transcriptionpolymerase chain reaction; CR, complete remission.ITo whom reprint requests should be addressed at: Box 305, Me-morial Sloan-Kettering Cancer Center, 1275 York Avenue, NewYork, NY 10021.

2694

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 21

, 202

0

Proc. Natl. Acad. Sci. USA 89 (1992) 2695

This was radiolabeled with [a-32P]dCTP using commercialrandom-priming kits.RT-PCR Assay for PML/RAR-a Expression. Patients di-

agnosed with APL (M3) by morphologic criteria (19) wereentered into a clinical trial investigating RA as single-agenttherapy. Patients or their guardians gave written informedconsent, and the study was reviewed and approved in ad-vance by the center's institutional review board. Karyotypeswere determined on unstimulated bone marrow cultures after48 hr with the use of Giemsa- or quinacrine-banding tech-niques (20). Leukemic cells from APL patients were exam-ined for the presence of the PML/RAR-a fusion mRNA.Total RNA from bone marrow aspirates was subjected to RTwith random hexamers and Moloney murine leukemia virusreverse transcriptase followed by 40-50 cycles of PCR am-plification (GeneAmp 9600, Perkin-Elmer/Cetus) (21). Max-imum sensitivity was obtained without nonspecific amplifi-cation using 50 cycles (data not shown). Primers for RAR-awere synthetic 20-nucleotide oligomers at bases 235 (5'primer) and 404 (3' primer) of the cloned RAR-a cDNA (18).Primer pair II for PML/RAR-a comprised a 5' 20-nucleotideoligomer from the cloned PML cDNA (base 1074) (7) and the3' 20-nucleotide primer from RAR-a (base 404) (18). Primerpair I comprised flanking 20-mers at base 997 of PML andbase 462 of RAR-a. Primer III was a 20-mer spanning bases1482-1501 of PML (7). PCR products were electrophoresedon 2% NuSieve/1% SeaKem agarose (FMC) gels in a Trisborate/Na2-EDTA buffer and visualized by staining withethidium bromide. The specificity of PCR amplification wasalso examined by Southern blotting the amplified cDNA withhybridization to 32P-labeled probes for a 1000-bp EcoRI/KpnI fragment of the PML cDNA or the 640-bp EcoRI/Sst Ifragment of the RAR-a cDNA (7, 18). A control for theintegrity of RNA isolated from the bone marrow mononu-clear cells was the simultaneous RT-PCR amplification of theRAR-a mRNA.PCR amplification by the nested primer technique was

performed by using primer pair II for 25 cycles followed byprimer pair I for 25 cycles as described (22). To evaluate thesensitivity of this assay in leukemic cells, NB4 cells express-ing the PML/RAR-a transcript (ref. 23; a gift of M. Lanotte)

BType A

were serially diluted in HL-60 cells, which do not express thistranscript (5, 24). RNA was extracted as described above,and RT-PCR amplification was performed both with primerpair I and by the nested primer technique.DNA Sequence Analysis. Sequencing was performed by the

chain-termination sequencing reaction (25) applied to double-stranded cDNA PCR products cloned into the TA cloningvector (Invitrogen, San Diego). The predominant clonedamplified fragments were sequenced by incorporation ofadenosine 5'-[t_[35S]thio]triphosphate using T7 DNA poly-merase (United States Biochemical) (25). PCR primer pair Ifor the breakpoint region of PML/RAR-a provided thesequencing primers. Size fractionation of these DNA prod-ucts was performed with a 5% polyacrylamide/7 M urea gel.Autoradiography was performed with XAR film (EastmanKodak).

RESULTSRT-PCR Assay. Total cellular RNA was prepared from

bone marrow mononuclear cells ofpatients with morphologicAPL. As reported (5), Northern analysis revealed two pre-dominant patterns of aberrant RAR-a mRNA expression.Representative expression patterns are shown in Fig. 1A.PML/RAR-a cDNAs cloned from APL cells of a patient witheach pattern have the structures and reported breakpointsequences depicted in Fig. 1B (7, 8). cDNAs were preparedfrom bone marrow mononuclear cells of other APL patients.These cDNAs were isolated from APL cells expressing eitheraberrant mRNA pattern. A 20-bp oligonucleotide primer 5' toboth reported cDNA breakpoints in PML (7, 8) and a primer3' to the breakpoint in RAR-a (primer pair II in Fig. 1B) weresynthesized as described above. Fifty cycles of PCR ampli-fication produced the specific DNA products shown in Fig.1C. Total RNA from two patients with pattern A on Northernanalysis yielded the 445-bp PCR product shown in lanes 1 and2. Total RNA from patients expressing three bands onNorthern blotting, type B in Fig. 1 A and B, yielded themultiple RT-PCR products shown in Fig. 1C (lanes 3 and 4).RT-PCR of control RNA from nonpromyelocytic AML pa-tient samples, HL-60 cells, which do not contain the t(15;17)

125634- -

t l;

. MM//////////MAgrZZMA 3'1263 17374- -

PML = RARc.

PMULRARlx Type A CAGGGGAAAGCCATTGAGACC

PMURARct Type B G0GGAGGCAGCCATTGAGACC

C1353 -1078 -872 -

603 -

310 -281 -271 -

1 2 3 4 5 6 7 8 9 10 11 12

FIG. 1. (A) Northern analysis mRNA expression patterns for RAR-a in APL. Ten micrograms of total cellular RNA isolated from anonhematopoietic tumor expressing the normal RAR-a species (lane 1) and bone marrow mononuclear cells from two APL patients (lanes 2and 3) were hybridized to a random-primed 600-bp Pst I RAR-a cDNA probe. Arrows depict the position of the two normal sized RAR-a mRNAspecies, which are expressed in all three lanes. Lanes 2 (type A) and 3 (type B) show the two patterns of abnormal RAR-a expression seen usingthis radiolabeled cDNA probe. The positions of the 18S and 28S ribosomal bands are shown. (B) Schematic diagram and nucleic acid junctionsequence of the predominant RT-PCR-amplified PML/RAR-a breakpoint cDNA species in APL. The cDNA was obtained from mononuclearbone marrow cells of APL patients expressing each of the two abnormal mRNA patterns shown in A. Horizontal arrows I, II, and III indicatethe location of oligonucleotide primers used for PCR amplification ofPML/RAR-a cDNA. Vertical arrows depict the position in the PML exonicsequence of the breakpoint. Open triangle depicts the position of the reported alternative splicing site present in the PML cDNA (7, 8). Nucleicacid sequence contributed by RAR-a is underlined. (C) Application of the RT-PCR assay for the PML/RAR-a fusion mRNA to bone marrowmononuclear cells isolated from APL patients. The RT-generated cDNA from APL bone marrow samples or from control cells was amplifiedby using primer pair II (lanes 1-6) or primer pair I (lanes 7-12). The RT-PCR products were electrophoresed as described and were visualizedby staining with ethidium bromide. Lanes: 1 and 2, two APL cases expressing pattern A; 3 and 4, two APL cases expressing pattern B; 5 and6, non-APL controls. In lanes 7-12, the cDNA used in lanes 1-6 was amplified with an internal primer pair. Molecular size standards (bp) appearon the left.

Type B

- 28S

A

*;

1 2 3

-i88

Medical Sciences: Miller et al.

Ili ill--D _I. v -r>

Dow

nloa

ded

by g

uest

on

Nov

embe

r 21

, 202

0

26% Medical Sciences: Miller et al.

(24), and other human malignant cells did not yield amplifiedproducts for the PML/RAR-a transcript (Fig. 1C lanes 5 and6; data not shown).To evaluate the specificity of the multiple bands in Fig. 1C,

PCR amplification was performed with primer pair I, whichhybridizes internal to the sequence amplified by primer pairII (see Fig. 1B). As shown in Fig. 1C the PCR amplified bandsin lanes 7-10 are similar to those in lanes 1-4 except that eachspecies is decreased in size by 131 bp. To confirm that theseethidium bromide-stained bands were specific RT-PCR am-plification products of the PML/RAR-a transcript, the am-plified cDNA was blotted and hybridized to PML- andRAR-a-specific probes. Southern analysis was performed onthe size-fractionated RT-PCR products depicted in Fig. 1C byusing appropriate radiolabeled PML or RAR-a probes. Au-toradiography identified all the species visualized by ethid-ium bromide staining (data not shown).

Similar multiple bands were found in the RT-PCR amplifiedRNA from APL cells of other patients expressing pattern B.Autoradiography of Southern blot hybridizations of theseRT-PCR products to PML-specific probes revealed up to sixdistinct bands (data not shown). This RT-PCR finding mightarise from the presence of an alternative splicing site in thePML sequence included in the fusion mRNA of APL cellsexpressing pattern B, but not in cells expressing pattern A.RT-PCR was performed on RNA from leukemic cells of arepresentative patient with each expression pattern. This wasaccomplished with a 5' primer from the PML sequence 3' tothe reported alternative splicing site (labeled III in Fig. 1B)(7, 8). Fig. 2 shows that the leukemic cells of a patient withPML/RAR-a pattern A yields one RT-PCR amplified band of445 bp with primer pair II (lane 1), 314 bp with the internalprimer pair I (lane 2), and no amplified product with primerIII (lane 3). This is consistent with the location of primer III3' to the breakpoint of PML/RAR-a of type A. RT-PCRamplification ofRNA from the APL cells ofa patient with thetype B breakpoint yields multiple bands with primer pair II(lane 4) and primer pair I (lane 5). With primer III, thereported alternative splicing site is not included in the am-plified PML sequence, and only one band is seen (lane 6).These data suggest that the two patterns seen by Northern

analysis or RT-PCR amplification correspond to the tworeported breakpoints for the PML/RAR-a fusion mRNAdepicted in Fig. 1B. The predominant RT-PCR amplificationproducts isolated from the APL cellular RNA of threepatients with each pattern were sequenced and the findingsare summarized in Fig. 1B. For leukemic cells from each

1353-1078-872 -

603-

310-281 -

1 2 3 4 5 6

FIG. 2. Loss of alternatively spliced RT-PCR species by ampli-fication with progressive 3' PML primers. The RT-PCR productswere size-fractionated as described. The cDNA from APL bonemarrow mononuclear cells isolated from a patient with expressionpattern A (lanes 1-3) and with expression pattern B (lanes 4-6) wasamplified with primer pairs II (lanes 1 and 4) and I (lanes 2 and 5) orPML primer III with RAR-a primer II (lanes 3 and 6). Molecular sizestandards (bp) appear on the left.

patient expressing pattern A, the PML breakpoint occurredat base 1263 (7), and the RAR-a breakpoint occurred at base281 (18). The leukemic cells from three patients expressingpattern B had RAR-a breakpoints identical to those ofpatternA but had all PML breakpoints at base 1737 (7).

Assay Sensitivity. Dilutions of plasmid DNA containingPML/RAR-a cDNA as an insert were amplified to assess thesensitivity of PCR amplification. PCR amplification withprimer pair I for 50 cycles of as little as 300 molecules of thisPML/RAR-a cDNA (7) yielded a 314-bp PML/RAR-a banddetectable by hybridization to a PML-specific oligonucleo-tide probe (data not shown). The RT-PCR assay performedon RNA isolated from leukemic bone marrow aspirates ofpatients expressing pattern A required only 50 pg ofRNA toyield a detectable 314-bp PML/RAR-a band. SuccessfulRT-PCR amplification of RNA isolated from leukemic bonemarrow aspirates of patients expressing pattern B required0.5-10 ng ofRNA to detect the larger PML/RAR-a species.The use of a nested primer PCR amplification techniqueincreased sensitivity for these larger PML/RAR-a species by10- to 100-fold. Using the nested primer technique, RNAprepared from APL bone marrow mononuclear cells express-ing either reported PML/RAR-a transcript is RT-PCR am-plified with enhanced sensitivity. In experiments with clinicalAPL specimens, RT-PCR amplification of RNA for thenonrearranged RAR-a allele, which is coexpressed with thePML/RAR-a transcript, was a control for RNA integrity.Fifty picograms of cDNA prepared from intact total RNAwas routinely sufficient to detect the 169-bp amplified RAR-afragment. To explore the sensitivity of this assay to detectisolated leukemic cells, NB4 cells, which express PML/RAR-a (23), were mixed with increasing proportions of thenontranslocated HL-60 cells. As shown in Fig. 3, PML/RAR-a can be detected by ethidium bromide staining of thecDNA produced from as little as 50 pg of NB4 total RNA.

Clinical Application of the RT-PCR Assay. Thirty-six pa-tients meeting clinical criteria for the diagnosis of APL wereentered into a trial investigating the antitumor activity ofRA.Leukemic cells isolated from these patients were examinedfor expression of the PML/RAR-a fusion mRNA by RT-PCR. As shown in Table 1, leukemic cells from 24 patientscontained a t(15;17) by cytogenetics. All 24 were also positivefor a PML/RAR-a transcript by RT-PCR, including 9 show-ing pattern A and 15 showing pattern B. In 4 patients withnondiagnostic cytogenetics, the RT-PCR assay was alsopositive. A normal bone marrow culture karyotype was foundin 8 patients. Application of the RT-PCR assay to the bonemarrow mononuclear cells purified from these 8 patientsyielded a positive result in 4 of them. Regardless of the

.- 3

7 ."

FIG. 3. Sensitivity of the RT-PCR assay applied to leukemiccells. NB4 cells, which express PML/RAR-a mRNA, were dilutedwith an increasing proportion of HL-60 cells, which do not expressthis transcript. Ethidium bromide-stained bands from RT-PCR am-plified species, which were size-fractionated as described in Fig. 1Clegend, are shown. These RT-PCR species were derived from cellulardilutions containing NB4 fractions of 1, 10-1, 10-2, 1i-0, 1O-4, and0. The depicted species were derived from 50 ng, 5 ng, 500 pg. 50 pg,5 pg, and 0 pg of NB4 total cellular RNA, respectively. Molecularsize standards (bp) appear on the left.

Proc. Natl. Acad Sci. USA 89 (1992)

AN ,,4 r ;-1 r -.1

Dow

nloa

ded

by g

uest

on

Nov

embe

r 21

, 202

0

Proc. Natl. Acad. Sci. USA 89 (1992) 2697

Table 1. Correlation of cytogenetic, Northern analysis, andRT-PCR assays for the rearranged RAR-a with the clinicalresponse to RA treatment in APL patients

Karyotype

t(15;17)NLNLNEM

t(15;17)t(15;17)t(15;17)NL

t(15;17)t(15;17)t(15;17)t(15;17)t(15;17)t(15;17)t(15;17)NLNEM

t(15;17)t(15;17)t(15;17)NEM

t(15;17)t(15;17)t(15;17)NLNL

t(15;17)t(15;17)t(15;17)t(15;17)t(15;17)NEMNL

t(15;17)t(15;17)NL

NorthernanalysisABANEBANE

AABBBBB

NEBB

PCR

ABABBAB

AABBBBB

BBB

A AB BNE BA AA AB BA AB BA AB BNE BA AB B

B B

B B

RA response

+

ED+

ED

+

ED

NL, normal; NEM, no evaluable metaphases; A or B, expressionpattern A or B (see Fig. 1A legend); NE, not evaluable (indicatingtotal RNA yields that were insufficient for Northern analysis); -, noabnormal RAR-a mRNA (Northern), no RT-PCR amplified PML/RAR-a (PCR), or no response to RA (RA response); ED, early death;+, clinical response to RA therapy.

cytogenetic findings, identification of a PML/RAR-a fusionmRNA by the RT-PCR assay predicted the CR induced byRA in all clinically evaluable patients.

Table 1 correlates the pattern seen on Northern analysiswith that seen by the RT-PCR assay. Twenty-seven of 31patients with adequate leukemic cell RNA for Northernanalysis showed one of the two patterns (Fig. 1 A and B) ofaberrant RAR-a expression. These patients were all positiveby RT-PCR, with the amplified cDNA pattern matching thepattern seen on Northern analysis. The leukemic cells from4 patients exhibiting only normal RAR-a expression onNorthern analysis were also negative by RT-PCR. In 5 cases,Northern analyses performed on mononuclear bone marrowcells were not evaluable because of inadequate total RNAyields or nonintact RNA. In each of these patients, theRT-PCR assay was diagnostic for the PML/RAR-a tran-script, indicating the enhanced clinical sensitivity of thisassay.The RT-PCR assay was used serially in bone marrow

mononuclear cells from APL patients responding to RAtreatment. At the time of entry into CR defined by conven-

tional clinical and cytogenetic criteria, the RT-PCR remainedpositive in the bone marrow cells from all patients evaluated.Six of these patients remained on continuous RA treatmentand received no cytotoxic therapy. All of these patientsrelapsed within 3-10 months. The subsequent bone marrowmononuclear samples yielded RNA and cDNA of variablequality. The presence of the PML/RAR-a transcript wasdetermined in bone marrow specimens withRNA ofadequateintegrity to detect expression of the nontranslocated RAR-aspecies. Using the nested primer assay, PML/RAR-a wasdetected in all the evaluable specimens from 5/6 patients. Inthe patient with the most durable CR (10 months), a singlesample, obtained 90 days after RA treatment began, yieldeda negative result. Subsequent testing of the bone marrowmononuclear cells from this patient yielded a positive result6 months before clinical relapse.

This RT-PCR assay has also been applied to APL cellsfrom patients who achieved CR after cytotoxic chemother-apy. To date, bone marrow mononuclear cells from 8 patientsin CR for 1 month to 9 years have been examined. Leukemiccells from 3 of these 8 patients tested positive for thePML/RAR-a rearrangement by RT-PCR. Although clinicalfollow-up of these patients is short, one patient relapsed 1month after the positive RT-PCR assay.

DISCUSSIONSeveral reports have shown that RA is a safe and highlyeffective alternative to cytotoxic chemotherapy for inducingCR in APL patients (3-5, 26). We previously found thatabnormal RAR-a mRNA expression correlates with theability of RA to induce maturation of leukemic cells andclinical remission (5, 9). With the cloning of the PML/RAR-afusion cDNA (7, 8, 12), it is possible to design specificmolecular probes. Genomic DNA probes have been derivedto reveal DNA rearrangements in APL cells (11, 13, 14). Themore sensitive RT-PCR method identified PML/RAR-a tran-scripts in selected cases (7, 8).This study describes an RT-PCR assay that detects both

reported forms of the PML/RAR-a rearrangement with highsensitivity. This assay was applied to leukemic cells obtainedfrom 36 patients whose APL diagnosis was made by conven-tional clinical criteria. The results reveal that this RT-PCRassay for the PML/RAR-a fusion mRNA is more sensitivethan either cytogenetics or Northern analysis for the detec-tion of this molecular rearrangement. The cytogenetic diag-nosis ofAPL is limited by the need to grow cells in short-termcultures to obtain evaluable metaphase preparations. North-ern analysis, while more sensitive than cytogenetics (5),requires a large yield of intact RNA. This is a particularproblem in APL, where promyelocytes have a high contentof lysosomal enzymes and are often obtained from bonemarrow aspirates in patients with a marked coagulopathy (2).

In this study, all clinically evaluable patients whose leu-kemic cells were positive by RT-PCR responded to RA. Nopatient with a negative RT-PCR assay responded to RAtreatment. Leukemic cells from four patients had adequatemetaphase preparations and a normal karyotype but werepositive for the PML/RAR-a rearrangement by RT-PCR.These four may represent a subset ofAPL patients analogousto leukemic cells of patients lacking the t(9;22) of chronicmyelogenous leukemia but expressing the BCR/ABL fusionmRNA (27). The positive clinical response of these patientsto RA contrasts with the failure ofRA to induce remission inthe four patients who tested negative for the rearrangementby both cytogenetics and RT-PCR. These data strongly arguethat the RT-PCR assay is both sensitive and specific for anabnormality in RAR-a and PML that is linked to RA re-sponse. Thus, this assay provides a specific molecular diag-nosis of APL with important clinical implications.

Patient123456789101112131415161718192021222324252627282930313233343536

Medical Sciences: Miller et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 21

, 202

0

2698 Medical Sciences: Miller et al.

The two patterns of abnormal RAR-a expression seen onNorthern analysis correspond to the two patterns ofRT-PCRamplified products. In cells expressing the PML/RAR-atranscript of pattern B, multiple specific amplification prod-ucts were seen. These multiple forms may result from alter-native splicing in the PML portion of the fusion molecule.Consistent with this interpretation, the number of RT-PCRspecies decreased with the use of a more distal PML primer(primer III in Fig. 1B).These multiple amplified forms and the difficulty in puri-

fication of intact, high molecular weightRNA from APL cells(5) may account for the decreased sensitivity of the RT-PCRassay for RNA of pattern B. As reported in other systems(22), sensitivity was increased by use of nested primers. Withthe described RT-PCR assay, PML/RAR-a of either patterncould be detected from 50-500 pg of bone marrow mononu-clear cell RNA. This is somewhat higher than the 10 pg ofRNA required for detection of BCR/ABL fusion RNA inCML cells (28). Since the PCR assay for PML/RAR-a isquite sensitive for the amplification of plasmid DNA con-taining the PML/RAR-a cDNA, these differences in sensi-tivity might reflect lower levels ofPML/RAR-a expression inAPL cells than of BCR/ABL in CML cells.An important application for this sensitive and specific

assay forPML/RAR-a expression is the detection of minimalresidual disease. Despite the high CR rate to RA therapy,most patients relapse who are not given cytotoxic chemo-therapy after an RA-induced remission (26, 29). This RT-PCRassay detected the presence of APL leukemic cells duringclinical CR in six of six patients treated with RA alone. Onlyone RNA sample containing amplifiable normal RAR-amRNA (a control for RNA integrity) was negative by thenested RT-PCR assay for PML/RAR-a. This occurred inbone marrow cells from the single patient who remained incontinuous CR for 10 months. Perhaps in this patient with themost durable CR after RA treatment the number of PML/RAR-a-expressing cells fell transiently below the assay'ssensitivity. The assay was positive in only three of eightpatients in CR after cytotoxic chemotherapy. It is not yetclear whether RT-PCR positivity will predict relapse in thisgroup or in those who received consolidation chemotherapyafter a RA-induced remission. If this assay proves to be anaccurate predictor of relapse, it may become possible todistinguish subsets ofAPL patients who would benefit fromadditional therapy. Alternatively, cured patients could bespared the risks of additional chemotherapy treatment.

In summary, this RT-PCR assay for the PML/RAR-a fusionmRNA provides an important tool for the molecular diagnosisof APL. The assay is a more sensitive and specific predictorofclinical response to RA therapy than either light microscopyor cytogenetics. The RT-PCR assay also permits moleculardissection of the fusion transcript and correlation of its ex-pression pattern in bone marrow cells during treatment ofAPLpatients. Our initial findings show that this assay providescritical prognostic information through the detection of min-imal residual disease in the bone marrow aspirates of patientsin clinical CR after RA therapy. Ultimately, this assay mightbe useful to determine the amount and adequacy of postrem-ission therapy administered to APL patients.We are indebted to Ms. Denise Moy and Ms. Nikki Feirt for expert

technical assistance. We thank Dr. M. Lanotte for use ofthe NB4 cellline. R.M.E. is an Investigator of the Howard Hughes MedicalInstitute at the Salk Institute for Biological Studies. These studieswere supported in part by the Howard Hughes Medical Institute, theWeingart Foundation, the National Institutes of Health Grant 1R01-CA54494-01, the Mathers Foundation, Grant FD-R-000674 of theFood and Drug Administration, and Grant PDT-381 from the Amer-ican Cancer Society. E.D. was supported in part by the AmericanCancer Society Clinical Oncology Career Development Award 90-129. W.H.M. is a recipient of the Young Investigator Award of the

American Society of Clinical Oncology. S.R.F. was supported by theMortimer J. Lacher Research Fund and is a recipient ofPublic HealthService Cancer Chemotherapy Training Grant CA-09207-14.

1. Larson, R. A., Kondo, K., Vardiman, J. W., Butler, A. E.,Golomb, H. M. & Rowley, J. D. (1984) Am. J. Med. 76,827-841.

2. Rodeghiero, F., Avvisati, G., Castman, G., Barbui, Y. &Mandelli, F. (1990) Blood 75, 2112-2117.

3. Huang, M., Ye, Y., Chen, S., Chai, J., Lu, J., Lin, Z., Gu, L.& Wang, Z. (1988) Blood 72, 567-572.

4. Castaigne, S., Chomienne, C., Daniel, M. T., Ballerini, P.,Berger, R., Fenaux, P. & Degos, L. (1990) Blood 76, 1704-1709.

5. Warrell, R. P., Jr., Frankel, S., Miller, W. H., Jr., Scheinberg,D. A., Itri, L. M., Hittelman, W., Vyas, R., Andreeff, M.,Tafuri, A., Jakubowski, A., Gabrilove, J., Gordon, M. S. &Dmitrovsky, E. (1991) N. Engl. J. Med. 324, 1385-1393.

6. de The, H., Chomienne, C., Lanotte, M., Degos, L. & Dejean,A. (1990) Nature (London) 347, 558-561.

7. Kakizuka, A., Miller, W. H., Jr., Warrell, R. P., Jr., Frankel,S. R., Murty, V. V. V. S., Dmitrovsky, E. & Evans, R. M.(1991) Cell 66, 663-674.

8. de The, H., Lavau, C., Marchio, A., Chomienne, C., Degos, L.& Dejean, A. (1991) Cell 66, 675-684.

9. Miller, W. H., Jr., Warrell, R. P., Jr., Frankel, S. R.,Jakubowski, A., Gabrilove, J. L., Muindi, J. & Dmitrosky, E.(1990) J. Natl. Cancer Inst. 82, 1932-1933.

10. Chomienne, C., Ballerini, P., Balitrand, N., Huang, M. E.,Krawice, I., Castaigne, S., Fenaux, P., Tiollais, P., Dejean, P.,Degos, L. & de The, H. (1990) Leukemia 4, 802-807.

11. Longo, L., Pandolfi, P. P., Biondi, A., Rambaldi, A., Men-carelli, A., LoCoco, F., Diverio, D., Pegoraro, L., Avanzi, G.,Tabilio, A., Zangrilli, D., Alcalay, M., Donti, E., Grignani, F.& Pelicci, P. G. (1990) J. Exp. Med. 172, 1571-1575.

12. Pandolfi, P. P., Grignani, F., Alcalay, M., Mencarelli, A.,Biondi, A., LoCoco, F., Grignani, F. & Pelicci, P. G. (1991)Oncogene 7, 1285-1292.

13. Alcalay, M., Zangrilli, D., Pandolfi, P. P., Longo, L., Men-carelli, A., Giacomucci, A., Rocchi, M., Biondi, A., Rambaldi,A., LoCoco, F., Diverio, D., Donti, E., Grignani, F. & Pelicci,P. G. (1991) Proc. Natl. Acad. Sci. USA 88, 1977-1981.

14. Biondi, A., Rambaldi, A., Alcalay, M., Pandolfi, P. P., LoCoco, F., Diverio, D., Rossi, V., Mencarelli, A., Longo, L.,Zangrilli, D., Masera, G., Tiziano, B., Mandelli, F., Grignani,F. & Pelicci, P. G. (1991) Blood 77, 1418-1422.

15. Chirgwin, J. M., Przybyla, A. F., MacDonald, R. J. & Rutter,W. J. (1979) J. Biochem. 18, 5294-5299.

16. Southern, E. M. (1975) J. Mol. Biol. 98, 503-517.17. Petkovich, M., Brand, N. J., Krust, A. & Chambon, P. (1987)

Nature (London) 330, 444-450.18. Giguere, F. V., Ong, E. S., Segui, P. & Evans, R. M. (1987)

Nature (London) 330, 624-629.19. Bennett, J. M., Catovsky, D., Daniel, M. T., Flandrin, G.,

Galton, D. A. G., Gralnick, H. R. & Sultan, C. (1985) Ann.Inter. Med. 103, 620-625.

20. Mohamed, A. N., Clarkson, B. D., Chaganti, R. S. K. (1986)Cancer Genet. Cytogenet. 20, 209-222.

21. Noonan, K. E. & Roninson, I. B. (1988) Nucleic Acids Res. 16,10366.

22. Hughes, T. P., Morgan, G. J., Mariat, P. & Goldman, J. M.(1991) Blood 77, 874-878.

23. Lanotte, M., Martin-Thouvenin, V., Najman, S., Ballerini, P.,Valensi, F. & Berger, G. (1991) Blood 77, 1080-1086.

24. Dalton, W. T., Jr., Ahearn, M. J., McCredie, K., Freireich,E. J., Stass, S. A. & Trujillo, J. M. (1988) Blood 71, 242-247.

25. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.Acad. Sci. USA 74, 5463-5467.

26. Chen, Z.-X., Xue, Y.-Q., Zhang, R., Tao, R.-F., Xia, X.-M.,Li, C., Wei, W., Zu, W.-Y., Yao, X.-Z. & Ling, B.-J. (1991)Blood 78, 1413-1419.

27. Tkachuk, D. C., Westbrook, C. A., Andreeff, M., Donlon,T. A., Cleary, M. L., Suryanarayan, K., Homge, M., Redner,A., Gray, J. & Pinkel, D. (1990) Science 250, 559-562.

28. Kawasai, E. S., Clark, S. S., Coyne, M. Y., Smith, S. D.,Champlin, R., Witte, 0. N. & McCormick, F. P. (1988) Proc.Natl. Acad. Sci. USA 85, 5698-5702.

29. Wang, Z. Y., Sun, G. L., Lu, J. X., Gu, L. J., Huang, M. E.& Chen, S. R. (1990) J. Exp. Clin. Hematol. 32, 34-35.

Proc. Natl. Acad. Sci. USA 89 (1992)

Dow

nloa

ded

by g

uest

on

Nov

embe

r 21

, 202

0