Improved Quantitation of Minimal Residual Disease in Multiple Myeloma Using Real-Time Polymerase...

1998;58:3957-3964. Cancer Res Christopher J. Gerard, Katherine Olsson, Rajeev Ramanathan, et al. StandardsPlasmid-DNA Complementarity Determining Region IIIMyeloma Using Real-Time Polymerase Chain Reaction and Improved Quantitation of Minimal Residual Disease in Multiple

Updated version

http://cancerres.aacrjournals.org/content/58/17/3957

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

[email protected] at

To request permission to re-use all or part of this article, contact the AACR Publications

Research. on February 21, 2014. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from




http://cancerres.aacrjournals.org/cgi/alerts

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]




http://cancerres.aacrjournals.org/




[CANCER RESEARCH 58. 3957-3964. September 1. 1998]

Improved Quantitation of Minimal Residual Disease in Multiple Myeloma UsingReal-Time Polymerase Chain Reaction and Plasmid-DNA Complementarity

Determining Region III StandardsChristopher J. Gerard,1 Katherine Olsson, Rajeev Ramanathan, Chris Reading, and Elie G. Hanania2

SySlemix. Inc., Palo Allo. California 9430J

ABSTRACT

The complementarity determining region III of the rearranged iinimi-noglobulin heavy chain gene has been the target for tumor-specific PCRassays for the detection and follow-up of B-cell malignancies. Previously,these assays have relied on gel-based end point data collection methods

(i.e., band densitometry) and, thus, have provided at best a semiquantitative assessment of tumor levels. We show the development of a novel,real-time TaqMan PCR assay to quantitate residual multiple myelomacells in clinical samples after high-dose chemotherapy and autologousstem cell transplantation. We provide evidence that real-time PCR isreproducible, sensitive, and quantitative. In a 40-replicate PCR experiment targeting the ÃŸ-actingene, the coefficient of variation for threshold

cycle data was 1.6%, whereas it increased to 13.6% and 31 %, respectively,for end point fluorescence and gel densitometry. Moreover, in an experiment directly comparing standard curves obtained from band densitometry and threshold cycle data, the standard curve constructed fromthreshold cycle data had a multiple R2 value of 1.00 and demonstrated a

dynamic range >4 logs, compared with the 2-log linear range of gel

densitometry. Finally, we show that when a complementarity determiningregion Ill-specific PCR primer is used in conjunction with a consensus

primer for the immunoglobulin heavy chain joining gene, plasmid DNAcan be used as a readily available and effective substitute for clonalplasma-cell genomic DNA when preparing standards. By applying real

time PCR to the analysis of clinical samples, we are able to quantitatelevels of tumor involvement with unparalleled reproducibility and statistical confidence. Real-time PCR technology may well provide the accuracy

and reliability necessary for minimal residual disease detection to havereal prognostic significance.

INTRODUCTION

Multiple myeloma is characterized by the clonal expansion ofplasma cells in the BM/ producing high amounts of monoclonal IgGor IgA. The Ig rearrangement that brings together the heavy chain VH,diversity (DH), and JMgenes results in a region of hypervariability, theCDR3, and represents a unique DNA sequence for each myelomaclone (1). Naturally, the sequence of this DNA rearrangement hasbeen the target for many variations of tumor-specific PCR assays forthe detection and follow-up of minimal residual disease for B-cell

malignancies (2).A crude PCR technique for detecting the presence of a clonal

population of B-cells is to use JH and VH consensus primers to look

for evidence of a single predominant PCR product (3, 4); multiplebands or a "smear" as detected with agarose or PAGE indicate a

Received 3/12/98; accepled 6/26/98.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore he hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1Current address: diaDexus LLC, 3303 Octavius Drive. Santa Clara. CA 95054.

Phone: (408) 330-5056; E-mail: [email protected] To whom requests for reprints should be addressed, at SyStemix. Inc., 1651 Page Mill

Road, Palo Alto. CA 94304. Phone: (650) 813-4124; Fax: (650) 856-4919; E-mail:[email protected].

1The abbreviations used are: BM, bone marrow; MRD. minimal residual disease:

CDR3, complementarity determining region III: Ig, Immunoglobulin: VH. immunoglobulin heavy chain variable gene; JH, immunoglobulin heavy chain joining gene: PB.peripheral blood: PB-HSC. peripheral blood hematopoietic stem cell: PBL, peripheralblood lymphocyte; Ct, threshold cycle.

polyclonal population of B-cells. Although simple, this assay is gen

erally not able to detect a clonal population of cells unless it represents>1% of the total cells in the reaction (5).

More typically, JH and Vlf consensus primers are used to screen formyeloma clones in heavily infiltrated BM (5, 6) or flow-cytometry-

sorted plasma cells (7). Once a unique clone is identified, the CDR3region is sequenced and can be used for designing an allele-specific

PCR primer (5, 8, 9). Generally, these CDR3 primers must be used inconjunction with a JH consensus primer to amplify the CDR3 sequence. The small size (<50 bp; Ref. 10) of this region precludes theuse of both 5' and 3' allele-specific PCR primers. An alternative

technique is to design a hybridization probe specific to the CDR3sequence, which is then used in a Southern format ( 11) to detect thetumor-specific rearrangement in a V,,/JH consensus PCR reaction(12-15).

Some attempts have been made to develop quantitative CDR3-

specific PCR assays, which rely on end point data collection. Forexample, the amount of CDR3-specific product formation for a set of

standards (as measured using band densitometry) can be correlatedwith known copy numbers of the target gene (5-7). From this data,

standard curves can be constructed to estimate the number of targetcopies in an unknown sample using the process of regression discrimination (16). The limiting dilution PCR assay (17, 18) is an alternativeand increasingly popular technique that does not rely on a standardcurve. For this analysis. Poisson statistics are applied to PCR-positive

serial dilutions of absolute cell numbers to predict the number oftarget copies in an unknown sample. However, both of these techniques are prone to wide variations in PCR product accumulation,which result in greatly expanded confidence intervals for the percentage of residual disease. Consequently, traditional end point measuresof PCR product accumulation are of limited quantitative potential.

Recently, the 5' nuclease assay has been described as a unique

detection system for PCR product accumulation (19). During theextension phase of the PCR, the 5'â€”Â»3'exonuclease activity of Taq

DNA polymerase cleaves a doubly labeled fluorogenic probe (20).These nonextendible probes (TaqMan probes; Roche Molecular Systems, Inc.), designed to hybridize internally to the PCR primers, arelabeled at the 5' end with a reporter dye (i.e., FAM, 6-carboxyfluo-rescein) and at the 3' end with a quencher dye (TAMRA, 6-carboxy-

tetramethyl-rhodamine). Fluorescence of the intact probe is quenchedmainly by FÃ¶rster-type energy transfer (21), and the nuclease degra

dation of the probe results in an increase in reporter dye fluorescence(22). The increase in fluorescence can be proportional to the concentration of template in the PCR.

The TaqMan method of PCR product detection has contributed tothe development of extremely sensitive, accurate, and high-throughput assays for the detection of a number of DNA templates (23-25).

Now, with the introduction of analytical thermal cyclers that arecapable of monitoring probe fluorescence real-time (i.e., during the

course of the PCR), it is possible to develop truly homogenous andquantitative PCR assays (26, 27).

We provide evidence in this report of the significant advantage ofreal-time PCR. Specifically, we demonstrate that real-time PCR data

3957




QUANTITATIVE REAL-TIME PCR FOR MRD

has a high degree of reproducibility and provides a substantiallyincreased dynamic range. Furthermore, we show that plasmid DNA isa readily available and an effective substitute for clonal B-cellgenomic DNA in constructing patient-specific standard curves. Fi

nally, we describe how these techniques are used clinically to provideaccurate quuntitation of minimal residual disease in a patient transplanted with autologous PB-HSC.

MATERIALS AND METHODS

Patient Samples

Multiple myeloma patients were participants in a Phase I/II clinical trialassessing the safety and efficacy of a tumor purging process in PB-HSC

transplantation. The patients had received s 12 months chemotherapy at the

time of study enrollment and provided informed consent.Clinical PB samples for MRD analysis were obtained before and after cell

processing. BM samples were collected at the time of enrollment and at 42,I(X). 180. and 365 days after transplant. BM samples collected before mobilization were stained with CD38-FITC and CD45-PE antibodies (PharMingen.San Diego, CA) and enriched for CD38* +/CD45"" cells using How sorting

(Becton Dickinson). Genomic DNA was isolated using the QiaAmp Blood Kit(Qiagen. Inc.. Chatsworth, CA), and eluted in 50-200 /j.1of 10 mM Tris, pH 8.0

(BioWhittaker. Inc.. Walkersville. MD). DNA concentration was determinedusing the picoGreen dsDNA Quantitation Kit (Molecular Probes, Eugene. OR)and a luminescence spectrometer (model LS-50B; Perkin-Elmer Corp., Nor-

walk, CT).

Ig Heavy Chain Consensus PCR

Consensus PCRs contained 0.1-0.5 /ig of DNA in a total volume of 50 /Â¿I.

Each reaction consisted of PCR Buffer #10 (10 mM Tris pH 9.2, 1.5 mMMgCl,. 75 mM KCI: Stratagene. La Jolla. CA), 0.2 /IM of each dNTP (dATP.dGTP. dCTP, dTTP) in a premixed solution (Boehringer Mannheim Corporation), and 1.0 units of AmpliTaq DNA Polymerase (Perkin-Elmer Corp.).

Primer sequence and concentrations are shown in Table 1. Standards andcontrol reactions (described below) include a positive control of 0.1 /j,g ofgenomic DNA from the IM9 cell line (CCL-155; American Type Culture

Collection, Manassas, VA), a negative control of 0.1 ^g of genomic DNAfrom the RPMI 8226 cell line (CCL-155; American Type Culture Collection),and a no-DNA control. During reaction set-up, all reagents and DNA were kepton ice, and 0.2 ml of thin-walled PCR tubes (Applied Scientific. South SanFrancisco, CA) were stored in a metal TempBlock prechilled to -20Â°C (USA

Scientific. Ocala, FL). Samples were then transferred to the thermal cycler(model 9600, PE/ABI) once the initial ramping temperature exceeded 80Â°C.Thermal cycler parameters included 2 min at 96Â°C.40 cycles of (15 s at 95Â°C,15 s at 57Â°C,15 s at 72Â°C).5 min at 72Â°C,and a 4Â°Csoak.

Consensus PCR product was ligated directly into the pCR2.1 plasmid vectorand transformed into TOPI OF' competent EscherichÃ¬a coli cells as per the

protocol of the Original TA Cloning Kit (Invitrogen Corporation. San Diego,CA). CDR3 DNA inserts were sequenced in both directions by Seqwright LLC(Houston, TX).

Allele-specifÃ¬c CDR3 PCR

In allele-specific CDR3 PCRs (alternatively referred to as idiotype-specitic

amplification), all standards and unknowns were analyzed at least in duplicateand contained a total of 0.3 ^tg of DNA/50 fil of reaction (equivalent to 50.000cells at 6 pg of DNA/cell). PCR reaction components were supplied byPerkin-Elmer Corporation and included 10 x PCR Buffer (100 mM Tris pH8.3, 500 mM KCI), 2.5-3.0 mM MgCU, 0.2 /IM each of dATP, dCTP, dGTP,0.4 fjLMdUTP. 0.5 units of AmpErase Uracil N-Glycosylase, 1.5 units of

TaqGold DNA Polymerase. and sterile water (Baxter Healthcare Corporation.Deerfield. IL). Primers and TaqMan probes were synthesized by the OligoFactory (PE/ABI; Foster City. CA). and were used at the assay-specificconcentrations shown in Table 1. TaqMan probes were labeled at the 5' end

with the reporter dye molecule FAM (6-carboxy-fluorescein: emission Ã€518nm). and at the 3' end with the quencher dye molecule TAMRA (6-carboxy-

tetramethyl-rhodamine; emission A 582 nm). In addition, the 3' end of the

probe was phosphorylated to prevent extension of the probe during the PCR.Real-time PCR reactions were placed in 0.2 ml of MicroAmp optical tubes

with caps (Perkin-Elmer Applied Biosystems Division. Foster City. CA), and

data were collected using the ABI PRISM 7700 SDS analytical thermal cycler,as described below. Thermal cycler parameters included 2 min at 50Â°C.IO min95Â°C,and 40 cycles of [15 s at 95Â°C,30 s at a patient-specific annealingtemperature (60Â°C-62Â°C)].

Standards. PBL DNA extracted from normal donors was used as a negative control. The DNA was diluted with 10 mM Tris (pH 8.0) to a finalconcentration of 30 ng//xl. Standard A (100% positive) was made by adding4.5 pg of plasmid DNA (containing the patient-specific subcloned CDR3

insert) to 250 /j.1of negative control DNA. Then, standard A was diluted 1:5into negative control DNA to produce Standard B (20% positive). Standard Bwas serially diluted 1:10 into negative control DNA to yield Standards C-G(2%-0.0002%). Standard H consisted of 100% negative control DNA.

Real-Time versus End Point: Infra-assay Reproducibility

Forty replicates of a 3 ng of PBL DNA sample were analy/.ed in a real-timeÃŸ-actinPCR assay. PCRs were performed as described above for the allele-

specific CDR3 assays. Primer/probe concentrations and sequence data arefound in Table 1. Thermal cycler parameters included 2 min at 50Â°C,10 minat 95Â°C.and 40 cycles of (15 s at 95Â°C, 1 min at 60Â°C). Data collectionprotocols are described in the "gel densitometry" and "ARn" sections below.

CDR3-specific Standards: Plasmid DNA versus Genomic DNA

Two sets of standards were prepared. One consisted of genomic DNA fromthe IM9 multiple myeloma cell line. The other was made by subcloning theIM9 CDR3 consensus PCR product into the pCR2.1 sequencing vector, asdescribed below. Both are in a background of genomic DNA from the RPMI8226 cell line. The RPMI 8226 cell line lacks a heavy chain gene and,therefore, is an effective negative control for both the consensus PCR and theCDR3-specific PCR.

The genomic DNA standards A-G consisted of seven 5-fold serial dilutions

of 0.1 fig/fxl IM9 DNA into 0.1 pig/^l RPMI 8226 DNA to give finalconcentrations of 100%, 20%, 4%, 0.8%, 0.16%, 0.03%, and 0.006%. respectively. The final standard. H. consisted of only RPMI 8226 DNA. To construct

Table I PCR primers and TaqMun probes

OligoCons-JhCons-VhAS-

105TMP-105AS-

104TMP-104BAIBA2TMP-BAAS-1M9TMP-IM9Sequence

(5 '-3')"OrientationACCTGAGGAGACGGTGACC

(-)ACACGGC(CT)GT(GA)TATTACTGT

(+)ACTGGTTATCCCCCGGTCT(+)FAM-AGGGTTCCCTGGCCCCAGGAGTCA-TAMRA(-)GCATTTGGGAGCTACAAGGTGG

( +)FAM-TACCCTGGCCCCAGTGGTCGAACC-TAMRA(-)TCACCCACACTGTGCCCATCTACGA

(+)CAGCGGAACCGCTCATTGCCAATGG(-)FAM-ATGCCC(TAMRA)CCCCCATGCCATCCTGCGT(+)CAAAAGAAGGGGGGGTGACA(+)FAM-TGGCCCCAGATATCAAAAGGGTCAA-TAMRA

( - )Â¡M0.2-0.5*0.500.650.200.500.200.300.300.200.500.20DescriptionJoining

gene consensusprimerVariablegene consensusprimerPatient

105 CDR3primerPatient105 CDR3probe-Patient

104 CDR3primerPatient104 CDR3probeÃŸ-actingeneprimerÃŸ-actingeneprimerÃŸ-actingeneprobeIM9

CDR3primerIM9CDR3probe"

FAM. 6-carboxy-fluorescein; TAMRA.6-carboxy-tetraincthyl-rhodamine.'Concentrationof primer = 0.25 UM in consensus PCR; 0.5 Â¿IMin 104 assay; 0.2 Â¿IMin 105 assay.

3958





the plasmid DNA and genomic DNA standards with equal target copy numbers(i.e., CDR3 rearrangements), it was assumed that a single copy of the 4 kb ofpCR2.1 vector containing the cloned IM9 CDR3 region was equivalent to onecopy of the genomic CDR3 DNA sequence. Accordingly, it was determinedthat 0.5 pg of this plasmid DNA in a background of 1 /j.g of RPMI 8226genomic DNA would have equal numbers of CDR3 target copies as 1 fxg ofIM9 genomic DNA (i.e., both would be "100%" positive). To complete the full

set of standards based on copies of plasmid DNA, the 100%-positive sample

was serially diluted into RPMI 8226 DNA as per the genomic DNA standardsabove.

The validity of replacing genomic DNA standards with plasmid DNA wasassessed using either VH/JH consensus primers (see Table 1) or CDR3-specific/Jâ€ž primers. Both assays included an IM9 CDR3-specific TaqMan

probe and used the consensus PCR cycler protocol described above. AfterPCR, the samples were transferred to a white 96-well flat-bottomed plate(Perkin-Elmer Corp.). and fluorescence emission data were collected on theLS-50B luminescence spectrometer plate reader (Perkin-Elmer Corp.), as

described previously (24). Standard curves were constructed by plotting fluorescence intensity (ARQ) versus the log,0 of cell equivalents of target DNA.Probability values evaluating differences in plasmid DNA versus genomicDNA standard curves were generated in SYSTAT (SPSS. Chicago, IL) usinga multiple regression, dummy-coding technique (28).

Data Collection Methods

Gel Densitometry. After PCR, 8 ju.1of 6 x Tracking Dye (BioVenturesInc.) were pipetted into the 50-/xl of PCR reaction. Each reaction (15 pit) wasthen loaded onto a 3% NuSieve/1% SeaKem GTG Agarose gel (FMC Bio-

Products, Rockland. ME) in 1 X TBE buffer (BioWhittaker). Electrophoresiswas performed at 200 V for 30-45 min. Gel bands were visualized by staining

with cilmlium bromide (Life Technologies, Inc., Gaithersburg, MD), and thedigital image was captured using the Eagle Eye II Still Video System (Strat-agene). Band density was measured using One-Dscan software (Scanalytics,

Billerica, MA), and regression analysis was performed with SYSTAT software.

ARn Real-Time and End Point Data Collection. Real-time data were

collected using the ABI Prism 7700 SDS (Sequence Detection System, PEApplied Biosystems Division: Perkin-Elmer Corp.. Foster City. CA). TheSDS software calculated ARn using the equation ARn = (Rn +) -(Rn~),

where Rn+ equals the ratio of reporter and quencher dye at any given timeduring a reaction, and Rn" equals the ratio of reporter and quencher

baseline emissions during cycles 3-15. A fluorescent threshold was man

ually set across all samples in the experiment such that it bisected the

exponential phase of fluorescent signal increase (generally <0.1 arbitraryfluorescent units). The Ct was defined as the cycle number at which asample's ARn fluorescence crossed this threshold. As shown in Fig. 2, Ct

is linearly related to the log of starting copy number. Consequently,assay-specific standard curves of Ct versus log,,, of copy number were used

to estimate the copy number of DNA targets in unknown clinical samples.End point ARn (Fig. 1) was simply the fluorescence measured at the finalcycle of the PCR protocol.

RESULTS

Real-Time versus End Point: Intra-assay Reproducibility. oligonucleotide primers targeting a 295-bp region of the human ÃŸ-actin

gene were used in conjunction with a doubly labeled fluorescent probein a real-time PCR assay to investigate intra-assay variation betweenreal-time and two end point PCR measures (ARn, band density). The

Ct (Fig. 1A), end point fluorescence (ARn; Fig. 1/4). and band densitydata (Fig. IB) were collected for 40 replicates of a 3-ng PBL DNA

standard (500 cell equivalents), as described above.As shown in Fig. 1, the coefficient of variation, a direct measure of

the 40-replicate intra-assay variability, was lower for real-time Ct data

(1.6%) than for end point fluorescence data (13.6%) or band densitydata (31%). The potential outlier (*; Fig. 1, A and B) still crossed the

threshold at about the same cycle number as the other replicates,although it did not exhibit the same fluorescent rate of increase. Thisparticular PCR phenomenon, possibly due to late cycle inhibition, hasbeen noted by others (26), but does not significantly affect thevariability of real-time data.

Real-Time versus End Point: Standard Curves. Real-time PCR

Ct data were compared directly to end point data (band density) in aCDR3 allele-specific assay for patient 104. Fig. 2A shows the gel and

resulting standard curve obtained from end point band density data,whereas Fig. 2B depicts the threshold plots and the standard curvegenerated from real-time data. Both standard curves were fitted with

95% confidence intervals.The greatly expanded confidence bands in Fig. 1A are partly the

result of regressing only the 2-log linear portion of the data against the

log of cell equivalents (â€¢;range, between 1,000 and 10 cell equivalents). However, all six samples with a visible PCR band (Aâ€”Â»F,

BLJUJLJ.J

fÂ® 1.2

Â§ Â°-a

P 0.4

Endpoint^ 295 bpARn,

0 5 10 15 20 25 30 35 40

Cycle

EndpointAssayBandDensityARn3.13

1.47

0.98 0.2031.3% 13.6%Real-TimeCt25.38

0.40

1.6%

Average =

Std. Deviation =

Coeff. of Variation =

Fig. I. Primers targeting the human ÃŸ-actingene were used in conjunction with a TaqMan probe in a real-time PCR assay to investigate intra-assay variation between 40 replicatesof a 3-ng PBL DNA standard. A. real-time Ct and end point fluorescence (ARn) were measured using the 7700 Sequence Detection System software. B. after PCR, hand densities ofgel-separated samples were measured. A comparison of the coefficient of variation of all three PCR measures demonstrates that real-time Ct data has increased sample reproducibilityand reduced variation between replicates. *, a potential outlier among the 40 replicates.

3959





Standards

Fig. 2. A comparison between standard curves generated fromreal-lime and end point data. Plasmid DNA containing the cloned

CDR3 Ig heavy chain gene rearrangement for a multiple myelomapatient l()4 was serially diluted into normal PBLs such that Aâ€”Â»H

represent 50.000; lO.OtX): 1,000; 100; 10; 1; 0.1; and $ copies ofplasmiti DNA in a background of 50.000 PBLs: L. 100-hp DNAladder. .4, the posl-PCR samples were loaded onto an agarose gel,and hand densities were measured. Only the 2-log linear portion of

the plot (â€¢;range, between 1.000 and 10 cell equivalents) wastitled with a least-squares regression and 95*^ confidence intervals.lÃ¬,the threshold plot depicts ARn versus cycle number for the same-samples above. All six PCR-positive data points (Aâ€”Â»F;50,000copies to I copy) were used to generale (he log-linear regressionplot of Cl versus cell equivalents. The 95<7rconfidence intervals forthis standard curve are extremely narrow and the multiple R~ for the

regression equals 1.00.

LABCDEFGH

IS

.Â£â€¢g 10

â€¢¿�a

1m

B

1 0.1

Cell Equivalents

0 5 10 15 20 25 30 35 40

Cycle

200

.Â£ 25

<J

S 30

9F n

40

R2 = 1.00

Cell Equivalents

5(),(XX)copies to 1 copy) were used to generate the log-linear regression plot of the data collected real-time (Fig. 2ÃŸ).Therefore, this

figure not only demonstrates the strong linear relationship between theCt and the log of starting copy number (R2 = 1.00), but reveals the

>4-log dynamic range of the assay.

Plasmid DNA versus Genomic DNA Standards. The assay sensitivity and limit of detection were measured for both plasmid andgenomic DNA standards in two assays, one with VH/JH consensusprimers (Fig. 3/4) and the other with CDR3-specific/JH primers (Fig.

3S). Although both the variable and joining gene consensus primers(Cons-Vh and Cims-Jh in Table 1) showed 100% homology to the

plasmid DNA subcloned from the IM9 consensus PCR product, theyshowed only <95% and 90% homology, respectively, to the IM9genomic DNA sequence (29). The significant increase in PCR efficiency and limit of detection resulting from 100% matched versusmismatched primer/target DNAs can be seen in Fig. 3/4 (P = 0.001).

However, when at least one primer was CDR3-specific and, therefore,

100% homologous to the target DNA (Fig. 3ÃŸ),the sensitivity of bothgenomic and plasmid DNA standards was not significantly different(P = 0.2).

Clinical Application for Real-Time PCR. The 98-bp multiple-myeloma-specific CDR3 sequence of patient 105 (Fig. 4/4) was am

plified using consensus PCR primers. The locations of the consensusprimers, CDR3 allele-specific primer, and fluorogenic TaqMan probe

are shown in Fig. 4B.The 100%-homologous CDR3 primer and probe were used with the

JH consensus primer in a series of quantitative real-time PCR assays

to analyze MRD for patient 105. A representative standard curve forthis patient-specific assay is shown in Fig. 5/4. Each point of the

standard curve was analyzed in duplicate, and the regression wasfitted with 95% confidence intervals. A strong linear relationship wasdemonstrated for >4 logs (R2 = 0.997; 50,000 to 10 cell equivalents).

BConsensus Primers CDR3-Specific/JH Primers

Standardsâ€¢¿�Plasmid DMA

â€¢¿�Genomic DNA

Fig. 3. Plasmid DNA versus genomic DNA for allcle-.specific standardcurves. Assay sensitivity was measured for both plasmid and genomicDNA standards in two assays, one with VH/JH consensus primers (A) andthe other with CDR.Vspecific primers (B).

100K 10K KÂ» W 1 100K 10K 1K

Cell Equivalents

w

3960





Fig. 4. Patient-specific CDR3 PCR. A. the 98-bpband was the result of amplifying patient 105's

enriched B-cell DNA with consensus PCR primersand contains the allele-specific CDR3 region. ( + ),IM9-positive control; (-), RPMI 8226 DNA-nega-tive control; Buf. PCR buffer control; L, 100-bp

DNA ladder. B, the CDR3 sequence data for patient105 is shown. Box, 5' region of the joining gene;

brackets, CDR3-specific primer location; under

line, region of TaqMan probe hybridization. Thevariable gene consensus primer (Cons-Vh) is directly 5' of the hypervariable CDR3 region.

Patient105

Controls(+)(-) Buf

SOD

ICOâ€”

50 â€”¿�

98 bp

B

Cons-VhACACGGCTGTGTATTACTGT

Con*-Jh

5TCACCGTCTCCTCAGGT

GGGAGAGAT[ATITTOACTOOTIATCCCCCO]GTCTCDR3 primer

TTGACTCCTGGGGCCAGGGAACCCT

The limit of detection for a single replicate, 10 tumor cells/50,000normal cells, was typical for this assay.

Fig. 55 compares the percentage of detectable CDR3-specificmyeloma in sequential clinical samples for patient 105. The datarepresent the mean (Â±SE) of completely independent experiments(N, in the column at far right equals the total number of experiments performed for a given sample). Within each experiment, allsamples were analyzed in at least duplicate (intra-assay coefficientof variation for estimated cell numbers 2:100 was typically <2%;data not shown). The PB-HSC samples (from days 1, 2, and 3 ofapheresis collection) were negative for any detectable residualdisease in all replicates available for analysis. Taking into consideration the limit of detection and the total number of cells analyzedfor each of the apheresis collections, the assay sensitivity for days1, 2, and 3 was estimated to be <0.0092%, 0.017%, and 0.0092%,respectively.

Fig. 5C is a notched-box-plot (30) of the data in Fig. 5B, except thatonly those data points derived from at least four experiments wereincluded. This plot is analogous to a nonparametric / test for between-group differences. The vertical lines at the notch of each box representthe data median, and the spiked tips of the boxes represent 95%confidence intervals. It may be concluded with 95% confidence thattwo population medians are different if the tips of the boxes do notoverlap. Fig. 5C shows that the 3.5% of residual disease detected inthe BM sampled after chemotherapy was significantly reduced by 2logs, to a mean tumor load of 0.04% for the 3 days of apheresis. Thepercentage of residual disease detected in follow-up BM sampled at42, 100, 180,and 365 days after transplant was 4.0%, 2.0%, 0.2%, and0.7%, respectively.

DISCUSSION

Although high-dose chemotherapy and stem-cell transplantationtogether provide an effective short-term treatment for multiple myeloma (31-33), they rarely result in permanent remission. Patientrelapse may be attributed to residual tumor cells present in the

autografi (as a result of insufficient purity; Refs. 1,18, 34, and 35) orchemotherapy-resistant myeloma cells remnant in the patient (36).Thus, quantifying the frequency of residual disease may be used as aprognostic factor and an indicator of a therapy's efficacy.

Real-Time PCR. This study presents a sensitive, specific, andquantitative means to detect MRD. Our approach, a modified, realtime PCR assay using TaqMan technology, has many advantages overtraditional end point measures. A primary strength of the assay is itssimplicity. For example, real-time PCR data are interpreted usingsimple regression analysis, where the degree of assay linearity isreflected in the multiple R2 statistic. For most real-time PCR applications, R2 5: 0.99 are typical, and R2 ^ 0.98 are usually indicative

of a poorly optimized system and/or problems with operator technique(i.e., pipetting inconsistencies). We have developed more than a dozenreal-time PCR assays (data not shown), and standard curves nevergenerate R2 < 0.99.

Furthermore, we have demonstrated that the dynamic range ofreal-time data can span s4 logs of target DNA concentration (Fig. 2B\Fig. 5/4). Although it is possible to shift the 2-log dynamic range of atraditional PCR assay by manipulating target concentration and/orcycle number (5), this requires time-consuming sample-specific optimizations and increases the potential for variability inherent in thepreparation of sample dilutions. In a study by Billadeau et al. (6), thepercentage of MRD in the PB of multiple myeloma patients varied byas much as 6 logs (32-0.00088%). Certainly, for such studies an assaywith a large dynamic range would be particularly convenient.

The single most important advantage of real-time PCR is its highdegree of reproducibility. Whereas it has been reported that end pointPCR measures based on limiting dilution analysis (18) or band den-sitometry (5) can vary by as much as 3-10-fold for the same DNAsample, we demonstrate herein that real-time PCR data, because it iscollected during the phase of exponential product formation for eachindividual sample, has a high degree of both intra-assay (Fig. 1) andinter-assay reproducibility (Fig. 5). In addition, we and others (26)have observed that real-time Ct data does not seem to be affected by

3961




A


25

O,

a> 30<_>>Â»Oâ€¢¿�Q

O

inOD

35

R2 = 0.997

1 10 100 1K 10K 100K

Cell Equivalents

Fig. 5. Quantitative analysts of residua] disease inclinical samples (multiple myeloma patient IOS). A.a representative standard curve with a regressionfitted with 95% confidence intervals. Each point hasbeen analyzed in duplicate (R2 = 0.997; n = 10). B,

data (log,,, of percentage of residual disease) represent the mean of completely independent experiments (error bars represent SE for interassay variation). Each data point (depicted as N. in the far rightcolumn) represents the mean of at least two replicates (intra-assay coefficient of variation for estimated cell numbers ^100 was typically <2%).NEC., none detected. C. a notched-box-plot, analogous to a nonparametric / test for between-groupdifferences, for samples with data derived from atleast four experiments. It may be concluded with95% confidence that two population medians aredifferent if the tips of the boxes do not overlap. *.significant between-group difference (P ^0.05).

BFollow-up BM

Post Transplant

DAY+365DAY+180DAY+100DAY+42

I DAY 3DAY2

IDAYII DAY3

DAY 2I DAY1

PB Post ChemotherapyBM/Post ChemotherapyBM/Plasma Cell Sorted

CD34+Thy+PB-HSC

ApheresisProduct

0.01 0.1 1.0 10 100

Percent CDR3-Specific Myeloma

Follow-up BMPost Transplant

DAY+365

DAY+180 -

DAY+100 -

DAY+42

DAY 3 -

DAY 2

DAY1 -

BM/Post Chemotherapy

ApheresisProduct

Exarrple: 95%confidence interval

0.01 0.1 1.0 10

Percent CDR3-Specific Myeloma

presumably anomalous, stochastic PCR amplifications that result inreduced product formation (Fig. 1).

Although not immediately obvious, there are several benefits toreal-time TaqMan PCR that are inherent in the technology. First, theassay is extremely rapid because post-PCR processing steps, such as

Southern transfers and probe hybridization, are unnecessary. All relevant data are, indeed, collected real-time during the course of atypical 2-h PCR cycler program; data analysis can be completed in

less than 10 min. Second, the majority of the assay, from reactionset-up to data collection and analysis, is a closed-tube system. Not

only does this feature reduce the risk of false positives resulting fromPCR product carry-over contamination, but it eliminates variation

from additional pipetting steps. Moreover, this assay is amenable to

enzymatic control for carry-over contamination, such as the dUTP/uracyl-N-glycosylase system (37) that we routinely use for all of our

quantitative PCR assays. Such a measure is of critical importance forany diagnostic assay that is optimized to detect single copy targets.Finally, the TaqMan PCR is highly specific for the gene target ofinterest. A positive fluorescent signal can only be generated from aPCR amplicon if both the primers and the fluorescent probe hybridizeto the template DNA. Therefore, as we have previously demonstratedwith other TaqMan assays (24), PCR artifacts such as nonspecificbands or primer-dimers, as detected on an ethidium-bromide-stained

agarose gel, are not TaqMan positive.Plasmid DNA versus Genomic DNA Standards. DNA isolated

from clonal-plasma cells must be used not only to identify the CDR3

3962





sequence, but, ideally, to generate standards for subsequent CDR3assays. Realistically, this may not be feasible. The frequency of tumorcells is generally quite low, especially in patients pretreated with highdoses of chemotherapy and, thus, isolating a significant number ofthese cells could require a large-scale BM harvest. Moreover, we have

found that using cell sorters to obtain pure plasma cells at best yieldspopulations only enriched in myeloma cells (4-34%, based on

TaqMan technology; data not shown). Consequently, the practicallimitations of acquiring sufficient tumor cells has forced some investigators to report a limit of detection for patient-specific assays, whichare based on standards made from cell-line DNA (38).

However, not only do thermal cycling protocols and target DNAconcentrations affect PCR dynamics, but so, too, does the degree ofprimer/template sequence homology. One cannot make an absolutequantitative assessment of MRD when there is a disparate degree ofprimer/template sequence similarity between assay standards and unknowns. To maintain sequence homology between assay standardsand patient samples, CDR3 sequences were cloned into a plasmidvector and used as standard templates. Although the consensus primers showed identical sequence homology to plasmid DNA but notgenomic DNA, it was determined that standard curves obtained fromgenomic and plasmid DNA were not significantly different when atleast one of the primers used was CDR3-specific (Fig. 3). Moreover,

generating standard curves from cloned DNA, compared with standards made from heavily infiltrated BM (3), could yield more accuratedata because the exact concentration of allele-specific DNA can be

reliably quantified. Therefore, by relying on cloned CDR3 sequencesto generate standard curves, the problems associated with using generic cell-line standards or isolating large numbers of pure, clonal,sample-specific tumor cells can be avoided.

Clinical Application for Real-Time PCR. Although detecting the

presence of MRD is a critical part of many transplantation protocols,the quantitation rather than detection of MRD may reveal relevantcorrelations with patient parameters and have important prognosticimplications. Indeed, there have been a number of traditional, endpoint PCR assays developed specifically to generate quantitativeMRD data. Some of these studies, whether for multiple myeloma orother B-cell malignancies, represent attempts to correlate MRD with

clinical outcome (15, 18, 36, 38, 39), and others seek to evaluate thetumor purging efficacy of various treatment protocols (7, 40).

Although some semiquantitative data have been generated for follow-up BM and PB samples (41), most studies have only reported the

presence or absence of MRD (15, 38). For example, in one of the firststudies to suggest that molecular detection of MRD can have prognostic significance, there was a clear association between disease-free

survival and the lack of detectable MRD in BM after transplantation(36). In this study, it was found that 25 of 35 patients with consistentlyPCR-positive BM had relapsed, whereas 8 of 41 patients with MRD

detected in only some BM samples relapsed. Despite the heterogeneity of tumor involvement in the BM, these data suggest a possiblecorrelation between the absolute level of residual tumor and frequencyof relapse.

Real-time quantitative PCR can provide the necessary precision to

not only correlate PCR negativity with clinical outcome but also toreveal the relationship between the patients' measured residual tumor

load over time, and their potential for relapse. For example, althoughthe dynamic range of data for the follow-up BM samples in Fig. 5C

is <1.5 logs and the confidence intervals for each time point are theresult of interassay and not intra-assay variation, small changes in the

percentage of MRD can be detected. Specifically, the precision ofreal-time PCR data revealed an increase in the percentage of detectable residual disease from 6 months (0.2%)-1 year (0.7%) after

transplant. This increase in BM tumor load is currently being evaluated for clinical significance.

Conclusion. The use of real-time quantitative PCR is not limited to

multiple myeloma. We have developed similar assays based on specific chromosomal translocations that are prevalent in other commonB-cell malignancies, including the BCL-2 major breakpoint regionand the BCL-1 major translocation cluster. These assays have an

additional advantage over the current CDR3 approach in that they aredisease-specific rather than patient-specific and can be universally

applied (manuscript in preparation).In this report, we show that the CDR3 real-time PCR assay clearly

provides an advantage over routine diagnostic approaches such asnested PCR and limiting dilution analysis for detecting MRD. Onceoptimized, it is a rapid, closed-tube system that requires no post-PCRmanipulation and fewer cells than other allele-specific assays. With its

reduced risk of contamination, increased sensitivity, and expandeddynamic range, this assay is capable of generating reliable and quantitative data for assessing patient MRD and may well provide the levelof accuracy necessary for such measures to have real prognosticsignificance.

REFERENCES

1. Okada. A., and All, F. W. The variable gene assembly mechanism. In: T. Honjo andF. W. All (eds.). Immunoglobulin Genes. Ed. 2. pp. 205-234. San Diego. CA:

Academic Press. 1995.2. Segal. G. H.. Jorgensen. T.. Masih, A. S.. and Braylan, R. C. Optimal primer selection

for clonality assessment by polymerase chain reaction analysis: I. Low grade B-celllymphoproliferative disorders of nonfollicular center cell type. Hum. Pathol., 25:1269-1275. 1994.

3. Lemoli, R.. Fortuna. A.. Motta. M. R., Rizzi. S.. Giudice. V., Nannelti. A.. Martinelli,G.. Cavo. M., Amabile. M.. Mangiami. S.. Fogli. M.. Conte, R.. and Tura. S.Concomitant mobilization of plasma cells and hematopoietic progenitors into peripheral blood of multiple myeloma patients: positive selection and transplantation ofenriched CD34+ cells to remove circulating tumor cells. Blood. 87: 1625-1634.

1996.4. Lopez, M.. Lemoine. F. M.. Firat. H., Fouillard. L., Laporte. J. P., Lesage, S., Isnard.

F.. Stachowiak. J., Ferrer-Le Coeur, F.. Morel. P.. Najman. A.. Douay. L.. and Gorin,N. C. Bone marrow versus peripheral blood progenitor cells CD34 selection inpatients with non-Hodgkin's tymphomas: different levels of tumor cell reduction.

Implications for aulografting. Blood, 90: 2830-2838, 1997.5. Billadeau. D.. Blackstadt. M.. Greipp, P., Kyle, R. A.. Oken. M. M.. Kay. N.. and Van

Ness, B. Analysis of B-lymphoid malignancies using allele-specific polymerase chain

reaction: a technique for sequential quantitation of residual disease. Blood. 78:3021-3029, 1991.

6. Billadeau. D., Quam. L.. Thomas, W.. Kay, N.. Greipp. P.. Kyle. R.. Oken, M. M..and Van Ness. B. Detection and quantitation of malignant cells in (he peripheral bloodof multiple myeloma patients. Blood, 80: 1818-1824, 1992.

7. Gaziti. Y., Reading. C. C.. Hoffman, R.. Wickrema. A.. Vesole, D. H.. Jagannath. S.,Condino, J.. Lee. B.. Barlogie, B.. and Tricot, G. Purified CD34 + Lin-Thy+ stemcells do not contain clonal myeloma cells. Blood. 86: 381-389. 1995.

8. Jonsson. O. G., Kitchens. R. L.. Scott. F. C.. and Smith. R. G. Detection of minimalresidual disease in acute lymphoblastic leukemia using immunoglobulin hypervariable region specific oligonucleotide probes. Blood. 76: 2072-2079, 1990.

9. SchÃ¶lten.C.. Hilgarth, B.. Hilgarth, M.. Jaeger. U., Greinix, H. T., and Mannhalter.C. Predictive value of clone-specific CDR3 PCR in high-grade non-Hodgkin's

lymphoma: long-term persistence of a clone-specific rearrangement in a patient withsecondary high-grade NHL in remission after high-dose therapy. Br. J. Haematol.. 97:239-247, 1997.

10. Ichihara, Y.. Abe, M., Yasui. H., Matsuoka, H.. and Kurosawa, Y. At least five DMgenes of human immunoglobulin heavy chains are encoded in 9-kilobase DNAfragments. Eur. J. Immunol.. 18: 649-652, 1988.

11. Southern, E. M. Detection of specific sequences among DNA fragments separated bygel electrophoresis. J. Mol. Biol., 9Â«:503-517, 1975.

12. Yamada, M.. Hudson. S.. Tournay. O.. Bittenbender. S.. Shane, S. S.. Lange. B..Tsujimolo, Y.. CatÃ³n. A. J.. and Rovera. G. Detection of minimal disease in hematopoietic malignancies of the B-cell lineage by using third-complementarity-determining region (CDR-III)-specific probes. Proc. Nati. Acad. Sci. USA, 86:5123-5127, 1989.

13. Bersagel. P. L., Smith, A. M., Szczepek. A., Mam. M. J.. Belch. A. R.. and Pilarski,L. M. In multiple myeloma, clonotypic B lymphocytes are detectable among CDI9 +peripheral blood cells expressing CD38. CD56, and monotypic Ig light chain. Blood,85: 436-447, 1995.

14. Willems, P., Croockewit, A.. Raymakers. R.. Holdrinet. R.. Van Der Bosch. G.. Huys.E.. and Mensink. E. CD34 selections from myeloma peripheral blood cell autograftscontain residua] tumour cells due to impurity, not to CD34 + myeloma cells. Br. J.Haematol., 93: 613-622, 1996.

3963





15. Zwicky, C. S., Maddocks. A. B.. Andersen, N.. and Gribben. J. G. Eradication ofpolymerase chain reaclion deCectable immunoglobulin gene rearrangement in non-Hodgkin's lymphoma is associated with decreased relapse after autologous bone

marrow transplantation. Blood, 88: 3314-3322, 1996.16. Miller, R. G. Regression Techniques. In: Simultaneous Statistical Inference, Ed. 2,

Ch. 3, pp. 109-128. New York: Springer-Verlag. 1981.

17. Schiller, G.. Vescio, R., Freytes, C., Spitzer, G.. Sahebi, F.. Lee, M., Wu. C. H., Cao,J.. Lee, J. C., Hong, C. H.. Lichtenstein, A., Lili. M., Hall, J.. Berenson, R.. andBcrenson, J. Transplantation of CD34 + peripheral blood progenitor cells after high-dose chemotherapy for patients with advanced multiple myeloma. Blood. 86: 390-

397, 1995.18. Henry, J. M., Sykes, P. J.. Brisco. M. !.. To, L. B., and Juttner. C. A. Comparison of

myeloma cell contamination of bone marrow and peripheral blood stem cell harvests.Br. J. Haematol., 92: 614-619, 1996.

19. Holland, P. M., Abramson, R. D., Watson, R.. and Gelfand. D. H. Detection ofspecific polymerase chain reaction product by utilizing the 5'-3' exonuclea.se activity

of Thermus M/miiiius DNA polymerase. Proc. Nati. Acad. Sci. USA. 88: 7276-7280.

1991.20. Lee. L. G., Connell. C. R.. and Bloch. W. Allelic discrimination by nick-translation

PCR with fluorogenic probes. Nucleic Acids Res., 21: 3761-3766, 1993.21. FÃ¶rster,V. T. Zwischenmolekulare Encrgiewanderung und Fluoreszenz. Ann. Phys

ics, 2: 55-75. 1948.22. Livak. K. J., Flood, S. J. A., Marmara, J.. Giusti, W., and Deetz, K. Oligonucleotides

with fluorescent dyes at opposite ends provide a quenched probe system useful fordetecting PCR product and nucleic acid hybridization. PCR Methods Appi.. 4:357-362, 1995.

23. Bassler. H. A.. Flood, S. J. A.. Livak, K. J., Marmaro, J., Knorr, R., and Bait. C. A.Use of a fluorogenic probe in a PCR-based assay for the detection of Listerianumocytogenes. Appi. Environ. Microbio!.. 61: 3724-3728, 1995.

24. Gerard. C. J.. Arboleda, M. J.. Solar. G.. MulÃ©.J. J., and Kerr, W. G. A rapid andquantitative assay to estimate gene transfer into retrovirally transduced hematopoieticstem/progenitor cells using a 96-well formal PCR and fluorescent detection systemuniversal for MMLV-based proviruses. Hum. Gene Ther., 7: 343-354, 1996.

25. Wilham. P. K.. Yamashiro, C. T.. Livak, K. J., and Batt, C. A. A PCR-based assay forthe detection of Escherichia cali shiga-like toxin genes in ground beef. Appi. Environ.Microbiol., 62: 1347-1353. 1996.

26. Heid, C. A.. Stevens, J., Livak, K. J., and Williams, P. M. Real time quantitative PCR.Genome Res., 6: 986-994, 1996.

27. Gibson. U. E. M.. Heid. C. A., and Williams, P. M. A novel method for real timequantitative RT-PCR. Genome Res.. 6: 995-1001. 1996.

28. Dawson-Saunder. B., and Trapp, R. G. Statistical methods for multiple variables. In:Basic, and Clinical Biostatislics. Ch. 12. pp. 207-228. Norwalk. CT: Appleton and

Lange, 1990.

29. Watkins, B. A., Davis, A. E.. Fiorentini, S., and Reitz. M. S.. Jr. V-region and classspecific RT-PCR amplification of human immunoglobulin heavy and light chaingenes from B-cell lines. Scand. J. Immunol., 42: 442-448. 1995.

30. McGill. R.. Tukey. J. W.. and Larsen. W. A. Variations of box plots. Am. Stat.. 32:12-16, 1978.

31. Fermand, J. P., Chevret. S., Ravaud. P.. Divine, M., Leblond, V., Dreyfus, F.,Mariette. X.. and Brouet, J. C. High-dose chemotherapy and autologous blood stem

cell transplantation in multiple myeloma: results of a phase II trial involving 63patients. Blood, 7: 2005-2010. 1993.

32. Dreyfus, F.. Leblond V.. Ravaud. P.. Meile, J., Quarre, M. C.. Pillier. C.. Boccaccio,C., and Varet, B. Detection of malignant B cells in peripheral blood stem cellcollections after chemotherapy in patients with multiple myeloma. Bone MarrowTransplant., 75.- 707-711, 1995.

33. Bjorkstrand. B. Multiple myeloma, high-dose treatment and autologous stem celltransplantation: current status. Med. Oncol., 13: 23-30, 1996.

34. Gribben, J. G., Neuberg. D.. Barber, M., Moore, J., Pesek, K. W., Freedman, A. S..and Nadler, L. M. Detection of residual lymphoma cells by polymerase chain reactionin peripheral blood is significantly less predictive for relapse than detection in bonemarrow. Blood. 83: 3800-3807. 1994.

35. Brenner. M. K. Gene Marking. Hum. Gene Ther., 7: 1927-1936. 1996.36. Gribben, J. G., Neuberg, D., Freedman, A. S., Gimmi, C. D., Pesek. K. W.. Barber.

M.. Saporito. L., Woo. S. D., Coral. F., Spector, N., Rabinowe, S. N., Grossbard,M. L., Ritz, J., and Nadler, L. M. Detection by polymerase chain reaction of residualcells with the bcl-2 translocation is associated with increased risk of relapse afteraulologous bone marrow transplantation for B-cell lymphoma. Blood, 8/: 3449-

3457, 1993.37. Hartley, J. L., and Rashtchian. A. Dealing with contamination: enzymatic control of

carry-over contamination in PCR. PCR Methods Appi.. 3: S10-S14, 1993.38. Corradini. P.. Astolft. M.. Cherasco. C., Ladetto, M.. Voena, C.. Caracciolo, D..

Pileri, A., and Tarella, C. Molecular monitoring of minimal residual disease infollicular and mantle cell non-Hodgkin's lymphomas treated with high-dose chemo

therapy and peripheral blood progenitor cell autografting. Blood. 89: 724-731. 1997.

39. Gribben, J. G.. Freedman, A. S., Neuberg, D.. Roy. D. C.. Blake, K. W., Woo, S. D..Grossbard, M. L.. Rabinowe, S. N.. Coral, F., Freeman, G. J., Ritz. J., and Nadler.L. M. Immunologie purging of marrow assessed by PCR before autologous bonemarrow transplantation for B-cell lymphoma. N. Engl. J. Med.. 325: 1525-1533,

1991.40. Negrin. R. S.. Kiem. H-P.. Schmidt-Wolf, l. G. H.. Blume. K. G.. and Cleary. M. L.

Use of the polymerase chain reaction to monitor the effectiveness of ex vivo tumorcell purging. Blood, 77: 654-660. 1991.

41. Billadeau, D., Prosper. F.. Verfaillie. C., Wesdorf, D., and Van Ness. B. Sequentialanalysis of bone marrow and peripheral blood after stem cell transplant for myelomashows disparate tumor involvement. Leukemia (Baltimore), //: 1565-1570, 1997.

3964




Improved Quantitation of Minimal Residual Disease in Multiple Myeloma Using Real-Time Polymerase...

Documents

Transcript of Improved Quantitation of Minimal Residual Disease in Multiple Myeloma Using Real-Time Polymerase...