document

NATURE MEDICINE • VOLUME 5 • NUMBER 10 • OCTOBER 1999 1171

ARTICLES

The variable regions of the B-cell receptor immunoglobulin (Ig)heavy and light chains combine to form the unique antigen-recognition site of antibodies and contain determinants, calledidiotypes, that can themselves be recognized as antigens. The id-iotypic determinants of the immunoglobulin synthesized by aclonal B-cell cancer, such as follicular lymphoma (FL), are uniqueand can thus serve as tumor-specific antigens1, as initially demon-strated in mice2. Purified autologous immunoglobulin proteinhas been used as a vaccine for human patients with lymphoma3.Although isolated tumor regressions were found, that study3 wasnot designed to determine clinical efficacy, because most patientswere already in clinical remission, and standard tumor shrinkagecriteria could not be used. Instead, that study assessed immuno-genicity, mainly the induction of antibody responses in 50% ofpatients4.

However, T cells are essential for the eradication of tumor cellsin experimental model systems5,6. Given the combined needs offurther enhancing vaccine potency and priming for CD8+ T-cellresponses, a preclinical model was used to directly compare anumber of cytokines and experimental adjuvants for their abili-ties to enhance idiotypic specific tumor resistance7. That survey

demonstrated the superiority of granulocyte–monocyte colony-stimulating factor (GM-CSF), a choice also supported by earliergene therapy studies8.

FL is associated with a characteristic chromosome translocationthat brings the bcl-2 gene (chromosome 18) under the transcrip-tional influence of the immunoglobulin heavy chain gene (IgH;chromosome 14). Bcl-2/IgH translocations involving the majorbreakpoint region (MBR) can be used as a molecular marker forminimal residual disease using a very sensitive PCR technique9,10.Most FL patients in complete remission (CR) after conventionalchemotherapy still have tumor cells with t(14;18) detectable byPCR (ref. 10). Furthermore, patients with persistent circulatingtumor cells seem to be at increased risk of relapse. Here we evalu-ated an idiotypic protein vaccine, combined with GM-CSF, for itsability to exert anti-tumor effects as measured by the eliminationof cells with t(14;18) from the peripheral blood of uniformlytreated FL patients in first CR.

PCR optimizationAmong the 20 patients who achieved CR after chemotherapyand completed vaccination with the idiotypic determinant vac-

Complete molecular remissions induced by patient-specificvaccination plus granulocyte–monocyte colony-stimulating

factor against lymphoma

MAURIZIO BENDANDI1, CHRISTOPHER D. GOCKE4, CAROL B. KOBRIN5, FLOYD A. BENKO4,LARS A. STERNAS1, ROBIN PENNINGTON5, THELMA M. WATSON1, CRAIG W. REYNOLDS2,

BARRY L. GAUSE1, PATRICIA L. DUFFEY6, ELAINE S. JAFFE3, STEPHEN P. CREEKMORE2,DAN L. LONGO6 & LARRY W. KWAK1

1Department of Experimental Transplantation and Immunology, Medicine Branch, Division of Clinical Sciences,2Biological Resources Branch, Division of Cancer Treatment and

Diagnosis, and 3Laboratory of Pathology, National Cancer Institute,Bethesda, Maryland, USA

4Department of Pathology, Pennsylvania State University, Hershey, Pennsylvania, USA5Science Applications International Corporation, Frederick, Maryland, USA

6National Institute on Aging, NIH Gerontology Research Center, Baltimore, Maryland, USACorrespondence should be addressed to L.W.K.; email: [email protected]

Lymphomas express a tumor-specific antigen which can be targeted by cancer vaccination. Weevaluated the ability of a new idiotype protein vaccine formulation to eradicate residualt(14;18)+ lymphoma cells in 20 patients in a homogeneous, chemotherapy-induced first clini-cal complete remission. All 11 patients with detectable translocations in their primary tumorshad cells from the malignant clone detectable in their blood by PCR both at diagnosis andafter chemotherapy, despite being in complete remission. However, 8 of 11 patients con-verted to lacking cells in their blood from the malignant clone detectable by PCR after vacci-nation and sustained their molecular remissions. Tumor-specific cytotoxic CD8+ and CD4+ Tcells were uniformly found (19 of 20 patients), whereas antibodies were detected, but appar-ently were not required for molecular remission. Vaccination was thus associated with clear-ance of residual tumor cells from blood and long-term disease-free survival. Thedemonstration of molecular remissions, analysis of cytotoxic T lymphocytes against autolo-gous tumor targets, and addition of granulocyte–monocyte colony-stimulating factor to thevaccine formulation provide principles relevant to the design of future clinical trials of othercancer vaccines administered in a minimal residual disease setting.

© 1999 Nature America Inc. • http://medicine.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m

1172 NATURE MEDICINE • VOLUME 5 • NUMBER 10 • OCTOBER 1999

ARTICLES

cine, combined with GM-CSF (Table 1), 11 had tumors thatdemonstrated amplifiable MBR rearrangements and were thussuitable for molecular analysis of minimal residual disease. AllBcl-2/IgH rearrangements were sequenced (not shown).

To determine the sensitivity of the PCR in our hands, we usedDNA template from tumor biopsies, diluted into peripheralblood mononuclear cells (PBMCs) from a normal donor.Samples from all three patients (Fig. 1a), as well as titrations ofthe MBR-positive RL cell line (not shown), consistently showed asensitivity of detection of 1 malignant cell in 100,000. We nexttested whether outer and inner antisense primers spanning indi-vidual breakpoints might further improve sensitivity in samplesfrom five patients. However, when we substituted MBR-specificfor JH consensus primers, the consensus primers consistently am-plified pre-vaccination PBMC rearrangements with greater or atleast comparable sensitivity (Fig. 1b). Finally, we determined theoptimal source of DNA template. Initially, we analyzed both pre-vaccination bone marrow (BM) and PBMCs in parallel for seven

patients. All PBMC samples yielded unique rearrangements,whereas rearrangements were amplified from only three BMsamples (Fig. 1c). Given these results, we completed our analysisusing PBMCs.

PCR of serial minimal residual disease samplesAll assays were done by individuals (M.B., C.D.G. and F.A.B.)‘blinded’ to clinical information about the samples, in indepen-dent laboratories at two separate sites. Furthermore, we used amultiple-tube approach, considering any time point with at leasta single amplification of the clone-specific breakpoint to be PCR-positive11. At least 10 replicates of 1 µg DNA prepared fromPBMCs from each time point were assayed. When none of thereplicate amplifications was PCR-positive, the correspondingsample and time point was considered to be PCR-negative. Therewas nearly complete concurrence of results between replicates ofindividual samples (Table 2) and between the two laboratorysites; 175 of 182 (96%) individual PCR reactions yielded identical

results.All 11 patients had PCR-positive PBMCs at

the beginning of the study (beforechemotherapy), as well as before vaccina-tion, despite remaining in CR. However,eight patients converted to having PCR-neg-ative samples after vaccination (Fig. 2; threerepresentative patients). All eight patientshave remained PCR-negative in follow-upsamples for a median of 18+ months aftervaccination (Table 2; range, 8+ to 32+months). As an internal control, the β-actinpseudogene was amplified from all negativesamples. Furthermore, breakpoints amplifiedfrom pre-vaccine PBMCs were sequencedand found to be identical to those from theircorresponding tumors. The remaining threepatients have continuously had PCR-positivesamples (Fig. 2 and Table 2). Thus, at least 8of 11 (73%) patients cleared residual cellswith t(14;18) from the peripheral blood as aconsequence of vaccination.

Autologous tumor-specific T-cell cytokineresponsesWe assayed fresh or cryopreserved PBMCs

Fig. 1 PCR sensitivity and optimization. MW, molecular size markers (sizes,left margins); actin, β-actin pseudogene (amplified as internal control). a,Tumor biopsy cells from three patients (patient 36 results are shown) were di-luted into normal donor PBMCs. DNA template extracted from all dilutionsamples (above gel) was used in a nested PCR for bcl-2/IgH MBR amplifica-tion. The sensitivity of detection was consistently 1 cell in 100,000. b,Titrations of PBMC DNA into placenta DNA (patient 15 results are shown;

representative of three patients). Consensus primers amplified unique bcl-2rearrangements with greater or at least comparable sensitivity, comparedwith outer and inner antisense (3’) primers spanning individual breakpoints.c, Pre-vaccination PBMC (PB) and bone marrow (BM) samples, compared inparallel (patient 19 results are shown; representative of four patients). Allseven PBMC samples, compared with only three BM samples, yielded uniquebcl-2 rearrangements.

a b c

Table 1 Patient characteristics

OffPatient Age/sex Clinical Histologya therapyb GM-CSF Antibody Statusc

number stage (months) (mcg/m2) response (months)

4 55/F IV A FSC 12 100 + CCR (53+)6 35/M IV A FSC 6 100 - CCR (50+)7 59/M IV A FM 6 100 + CCR (50+)8 52/M IV A FSC 9 100 + CCR (50+)9 45M IV A FSC 8 500 + Relapse (28)

11 47M IV A FSC 12 500 - CCR (38+)12 62M IV A FM 9 500 + CCR (51+)14 48/M IV A FSC 10 100 - Relapse (22)15 54/F III A FM 15 500 - CCR (44+)16 28/M IV A FSC 10 100 + CCR (42+)18 45/F III A FM 15 100 + CCR (42+)19 58/F IV A FM 12 500 + CCR (43+)23 24/F IV A FM 9 500 + CCR (39+)25 48/F IV A FSC 15 500 + CCR (39+)27 45/M IV A FSC 12 500 + CCR (38+)28 26/F IV A FM 11 100 - CCR (37+)29 54/M IV A FM 12 100 + CCR (36+)30 38/M III A FSC 12 500 + CCR (39+)31 37/F IV B FSC 12 100 + CCR (36+)36 34/M IV A FSC 8 100 + CCR (28+)

aSubtype according to working formulation. bMonths between chemotherapy and vaccine treatment. cMonths aftercompletion of chemotherapy.


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m


ARTICLES

obtained after vaccination for specific cytokine release in re-sponse to autologous FL (Au FL) targets, available in each casefrom cryopreserved tumor biopsies, to carrier (keyhole limpethemocyanin) as a positive control, or to medium alone after asingle 6-day stimulation. There was considerable cytokine re-lease in response to Au FL targets in 19 of 20 (95%) patients.There were substantial levels of tumor necrosis factor (TNF; me-dian, 825 pg/ml; range, 92–2,453; Fig. 3a), as well as lower levelsof GM-CSF and gamma interferon (median, 112 pg/ml (range,53–372) and median, 121 pg/ml (range, 50–290), respectively;data not shown), compared with PBMCs or tumor alone.Moreover, PBMC cytokine release was dose-dependent in re-sponse to Au FL targets (not shown).

We did additional experiments using samples from six ran-domly selected patients (patients 6, 8, 15, 16, 19 and 23) to char-

acterize the cell-mediated responses. First, pre-vaccinationPBMCs assayed in parallel with post-vaccination PBMCs uni-formly failed to respond, indicating that these responses werevaccine-induced (Fig. 3b). In separate experiments, immunePBMCs from these patients were stimulated in parallel with AuFL targets, alone or with monoclonal antibodies against MHCclass I or II, as well as with autologous, non-neoplastic (normal)blood B cells as controls. TNF secretion was specific for autolo-gous tumor in all patients (median, 764 pg/ml; range 232–819),as demonstrated by the lack of significant cytokine productionby PBMCs stimulated with autologous, normal B cells in parallel(median, 0 pg/ml; range, 0–172); P = 0.003 by paired t-test; Fig.3c). Furthermore, addition of monoclonal antibody againstMHC class I resulted in significant inhibition of TNF production(median, 147 pg/ml; range, 34–229); P = 0.003), compared with

Fig. 2 Monitoring of minimal residual disease in the peripheralblood. Multiple replicates from each time point, each containing 1 µgPBMC genomic DNA template, were assayed simultaneously bynested PCR for bcl-2/IgH MBR translocations. Data represent productsfrom nested PCR of four to seven replicates each of PBMC DNA, tumor

lymph node (LN) DNA, or buffer alone from three of eight patientswho converted to having PCR-negative samples (a), and one of threepatients who continued to have PCR-positive samples (b). MW, mole-cular size markers (sizes, left margins); actin, β-actin pseudogene (am-plified as internal control).

a

b

Patient 28

Patient 15

Patient 7

Patient 9

Conversions to PCR negativity

Persistance of PCR positivity

Before vaccine

Before vaccine

Before vaccine

Before vaccine

After vaccine

After vaccine

After vaccine

After vaccine Actin

Actin

Actin

Follow-up (+15 mos)

Follow-up (+27 mos)

Follow-up (+17 mos)LN

Follow-up (+16 mos) LNDiagnosis

Diagnosis

Diagnosis

Diagnosis

Table 2 MBR PCR results for serial peripheral blood samples

Patient number 6 7 8 9 11 15 19 23 28 31 36

Timepoint

Study entry 10/10 10/10 10/10 10/10 10/10 10/10 10/10 10/10 10/10 10/10 10/10Pre-vaccine 10/10 7/14 10/10 10/10 10/10 12/17 10/10 10/10 10/10 10/10 10/10Post-vaccine (1) 0/10 0/14 0/10 10/10 10/10 0/17 10/10 0/10 0/10 0/10 0/10

Follow-up 0/10(7) 0/10(5) 0/10(20) 10/10(6) 10/10(6) 0/10(6) 10/10(11) 0/10(8) 0/10(8) 0/10(8) 0/10(9)

Follow-up 0/10(14) 0/10(13) 0/10(26) 10/10(15) 10/10(13) 0/10(11) 10/10(19) 0/10(18) 0/10(16)

Follow-up 0/10(25) 0/10(18) 10/10(19) 0/10(17)

Follow-up 0/10(32) 0/10(27)

Data represent the number of positive reactions/total reaction number. Numbers in parentheses, months after vaccine.


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m


ARTICLES

autologous FL targets with control monoclonal antibody (me-dian, 526 pg/ml; range, 191–783), whereas the addition of anti-body against MHC class II (median, 493 pg/ml; range, 208–752)was not associated with significant inhibition of cytokine pro-duction (P > 0.05). In another experiment, subsets of CD4+ andCD8+ T-cells selected from immune PBMCs were cultured withAu FL cells (Fig. 3d). TNF secretion by stimulated CD8+ T cells(median, 831 pg/ml; range, 565–2483) was comparable to that ofstimulated, unfractionated PBMCs (median, 1,173 pg/ml; range,742–2338). However, stimulated CD4+ T cells were also capableof producing low but significant amounts of TNF (median, 361pg/ml; range, 156–628) compared with that produced withmedium alone (0 pg/ml; P = 0.005). These results demonstratethat CD8+, MHC class I-restricted T cells specific for autologous FLcan be readily isolated from PBMCs of vaccinated patients.Furthermore, these T-cell responses were sustained for at least 6months after vaccination (range, 6+ to 18+ months).

We also stimulated immune PBMCs with purified, autologousFL immunoglobulin protein or a panel of control, isotype-matched FL immunoglobulins (50 µg/ml each). The same pa-

tients demonstrated a cytokine response after but not before vac-cination, specific for their unique idiotype (median, 1,296 pg/ml;range, 130–1,786), compared with the response using control, FLimmunoglobulin (median, 25 pg/ml; range, 15–96); P < 0.001,not shown). Moreover, the addition of monoclonal antibodyagainst MHC class II (median, 160 pg/ml; range, 127–345); P <0.001), but not against MHC class I (median, 764 pg/ml; range,243–1,245); P > 0.05), significantly reduced TNF secretion byPBMCs, compared with control monoclonal antibodies (median,1,459 pg/ml; range, 351–2,000) in the same six patients studied inFig. 3. Thus, in contrast to antigen presented on the surface of au-tologous tumor, stimulation with soluble tumor antigen alloweddetection of class II-restricted, antigen-specific T cells.

Lysis of Au FL targets by tumor-specific T cellsIn all six patients (patients 6, 8, 15, 16, 19 and 23), there was sub-stantial specific lysis of autologous, unmodified FL targets by im-mune PBMCs re-stimulated with a single 5-day cycle of CD40ligand (CD40L)-activated Au FL cells (range, 27–44%;effector:target (E:T) ratio, 100; Fig. 4a). Purified T cells positively

Fig. 3 Vaccine-induced, tumor-specific PBMC cytokine (TNF) responses.a, Post-vaccine PBMCs from all patients were cultured with either mediumalone, cryopreserved autologous tumor or autologous tumor alone.Supernatants were collected after 6 d and cytokine release was measured. *,single negative result. b, Pre- or post-vaccine PBMCs from six randomly se-lected patients, tested in parallel as in a and against KLH as a positive con-

trol. c, Post-vaccine PBMCs were co-cultured with Au FL or autologous, nor-mal B cells (Au nl B). FL cells were also pre-treated with monoclonal anti-body against MHC class I or II (α-MHC I or II) or control monoclonalantibody (Control Ab) before co-culture with PBMCs. d, CD4+ and CD8+ Tcells were positively selected from immune PBMCs by magnetic cell sortingand cultured with either medium alone or Au FL cells.

a b

c dPatient

Patient Patient

Patient

TFN

(p

g/m

l)

TFN

(p

g/m

l)TF

N (

pg

/ml)

TFN

(p

g/m

l)


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m


ARTICLES

selected from immune PBMCs (23–29% specific lysis; E:T ratio,20; not shown), but not pre-immune PBMCs (Fig. 4a), also lysedAu FL targets when similarly re-stimulated.

In additional experiments, this specific lysis (39–44%) wastumor-specific, as demonstrated by the inability to lyse labeledautologous Epstein-Barr virus lymphoblastoid lines and/or autol-ogous, normal B cells (< 1%; Fig. 4b). In addition, compared withcontrol monoclonal antibodies (31–37%), pretreatment of Au FLtargets with antibody against MHC class I (5–15%), but notagainst class II (20–28%), substantially reduced cytotoxicity.Finally, CD8+ subpopulations selected from stimulated PBMCswere each sufficient to mediate lysis of their respective unmodi-fied Au FL targets, even at lower E:T ratios (24–27%, E:T ratio, 20;Fig. 4c). Thus, CD8+, class I-restricted T cells, capable of lysingunmodified, Au FL targets, could be readily detected after, butnot before, vaccination.

Similar levels of lysis were obtained when highly purified AuFL cells (obtained by FACS after double staining with mono-clonal antibodies against the immunoglobulin heavy and lightchain isotype expressed by the patient’s tumor) were used as tar-gets (not shown).

Antibody responsesWe analyzed serum antibody responses by ELISA as described3.We detected post-vaccination responses specific for autologousidiotype, demonstrated by the lack of binding to a panel of con-trol, isotype-matched FL imunoglobulins, in 15 of 20 patients(Table 1); these were in all but one case mainly of the IgG1 sub-class (not shown).

Clinical courseEighteen of twenty patients remain in continuous, first CR (me-dian, 42+ months from completion of chemotherapy; range, 28+to 53+; Table 1). Patients 9 and 14 relapsed at 15 and 7 monthsafter vaccination, respectively. Patient 9 had never cleared thecells with t(14;18) from the peripheral blood; patient 14 did nothave the MBR rearrangement and thus, molecular CR statuscould not be established.

DiscussionOur experiments overcame two limitations of previous idiotypicdeterminant vaccine studies: the induction of tumor-specificCD8+ T cells and the demonstration of anti-tumor effects in a ho-mogeneous group of patients.

Although peptide-pulsed or allogeneic tumor cells have beenused as surrogate T-cell targets in other cancer vaccine stud-ies12–17, we chose the killing of autologous tumor targets as theminimum standard that should be achieved. Accordingly, ouranalysis using unmodified Au FL as targets of cytokine re-sponses and cytotoxicity after a single cycle of re-stimulationwith CD40L-activated tumor provides evidence for the genera-tion of tumor-specific, class I-restricted CD8+ T cells as a directresult of vaccination. Moreover, CD4+ T cells, which may be re-quired for the generation and maintenance of CD8+ cells, werealso elicited by vaccination (Fig. 3d and not shown). Our resultsmay provide the first demonstration that human FL cells are ca-pable of presenting endogenous immunoglobulin as a target ofCD8+ T-cells. Previous reports of FL-related cytotoxicity usedlymphoblastoid lines transfected with tumor immunoglobulingenes18 or a tumor hybridoma19 as targets, but not autologous FLcells. Similarly, active or adoptive immunotherapies targetingidiotype on other B-cell malignancies (such as myeloma20,21) will

Fig. 4 Cytotoxicity against unmodified, Au FL cells. After a single cycleof stimulation with CD40L pre-activated Au FL cells, pre- and post-im-mune PBMCs were recovered and used as effectors in a 12-hour indiumrelease CTL assay, against labeled unmodified, Au FL targets. Specificlysis of Au FL cells was mediated by post- but not pre-immune PBMCs(a); this lysis was tumor-specific, as shown by the lack of lysis of autolo-gous, normal B cells and autologous Epstein-Barr virus line targets, andwas blocked by monoclonal antibodies against MHC class I (b); and thelysis was mediated by CD8+ T cells (c).

a

c

b

Patient

Patient

Patient After Before

Effector: target ratio

Effector: target ratio

Spec

ific

lysi

s (%

)Sp

ecifi

c ly

sis

(%)

Specific lysis (%)

111 in

-lab

elle

d t

arg

ets


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m


ARTICLES

MethodsPatients and vaccine formulation. Thirty-five patients signed informedconsents for this investigational review board-approved study. Twenty-three (66%) patients achieved a clinical CR after chemotherapy. One ofthese patients was excluded because he relapsed 4 months after treatment,and two were excluded because vaccines could not be produced for techni-cal reasons, leaving a homogeneous group of 20 patients with FL in first CRwho received vaccine treatment (Table 1). Other patients found to haveresidual or progressive disease after chemotherapy who were administeredvaccines were excluded from this analysis, because the question of eradica-tion of minimal residual disease was not relevant.

Previously untreated patients with stage III/IV follicular small cleaved cellor mixed lymphoma confirmed by the Laboratory of Pathology, NationalCancer Institute (E.S.J.) underwent lymph node collection and were thenuniformly treated with six or more monthly cycles of prednisone, doxoru-bicin, cyclophosphamide and etoposide (ProMACE (ref. 27), withoutmethotrexate) to CR, plus two additional cycles. Documentation of CR afterchemotherapy and before and after vaccination required absence of diseaseon physical examination and bilateral bone marrow biopsies, and stabilityof residual abnormalities on computerized tomography scans and lymphan-giograms.

Immunoglobulin protein was isolated from each patient’s tumor by het-erohybridoma fusion and conjugated to keyhole limpet hemocyanin(Biomira, Edmonton, Canada) as described3. Identity of fusions and FL wasdetermined by comparing Ig VH CDR3 sequences28. After at least 6 monthsof immune recovery after chemotherapy, each patient remaining in CR re-ceived four monthly subcutaneous vaccinations with FL immunoglobulin(0.5 mg)–keyhole limpet hemocyanin mixed with 100 or 500 µg free GM-CSF per m2 body surface area (manufactured by Immunex, Seattle,Washington and provided by Cancer Treatment Evaluation Program,National Cancer Institute). For each course, three additional GM-CSF doseswere administered daily at the same site. An identical booster vaccinationwas given 2 months later.

DNA preparation and nested PCR assay. Genomic DNA was extractedusing the DNA Midi Kit (Qiagen, Santa Clarita, CA). Bcl-2/IgH genetranslocations were amplified by a nested PCR as described9, with minorchanges. Reactions containing 1 µg DNA (0.5 µg produced comparable re-sults) were first amplified for 30 cycles in 50-µl volume containing PCRBuffer II, 1.5 mM magnesium chloride, 200 nM oligonucleotide primers(Midland Certified Reagent Company, Midland, Texas), 200 nM each ofdATP, dCTP, dGTP, dTTP (Promega, Madison, Wisconsin), and 5 U AmpliTaq Gold. Unless otherwise specified, reagents were purchased from PerkinElmer (Norwalk, Connecticut). A 5-µl aliquot of the first-round reactionwas re-amplified for 30 cycles in a 50-µl reaction using internal primers.Primers for first- and second-round PCR were: 5’–CAGCCTTGAAA-CATTGATGG–3’ (MBR) and 5’–ACCTGAGGAGACGGTGACC–3’(JH, anti-sense); and 5’–TATGGTGGTTTGACCTTTA–3’ (MBR) and5’–ACCAGGGTCCCTTGGCCCCA–3’ (JH antisense), respectively. Aliquotsof the final product were separated by electrophoresis through 2% agarosegels (Life Technologies) containing ethidium bromide.

Lymphocyte proportions in mononuclear fractions of PBMCs did not varybetween pre-vaccine (mean ± s.e.m., 76% ± 4.2) and post-vaccine (mean ±s.e.m., 72% ± 3.8) samples. Serial pre- and post-vaccine samples were ana-lyzed in single experiments. For each time point, 10 or more replicates con-taining 1 µg DNA each were tested in parallel in the same experiment, withpositive (RL cell line), and negative controls (DHL-16 cell line and bufferalone). As an internal control, the β-actin pseudogene was amplified from allsamples, using the following primers: 5’–GTGGGGCGCCCCAGGCACCA–3’(sense) and 5’–CTCCTTAATGTCACGCACGATTTC–3’ (antisense).

Tumor-specific cytokine production. Pre- or post-vaccine PBMCs (2.5 x106) were cultured for 6 d with either medium alone (RPMI 1640 with 10%NCTC-109, 7% fetal bovine serum (Life Technologies), 2 mM glutamine, 10nM β-mercaptoethanol (Sigma), non-essential amino acids, 1 mM sodiumpyruvate, 100 U/ml penicillin + 100 µg/ml streptomycin), 5 × 104 cryopre-served, autologous tumor cells from lymph node biopsies (Au FL), or 50µg/ml keyhole limpet hemocyanin. Most cells in the lymph node biopsieswere tumor (range, 60–92%). Unless otherwise specified, reagents werepurchased from BioWhittaker (Walkersville, Maryland). Supernatants were

also require demonstration that the native tumor target ex-presses idiotypic peptide–MHC complexes on the surface.However, it is also possible that a tumor antigen unrelated to id-iotypic determinants is presented by FL cells for vaccine-in-duced T-cell recognition and lysis.

We chose rearranged bcl-2 instead of another molecularmarker, Ig VH CDR3, for this systematic analysis of eradication ofminimal residual disease because of potentially false negative re-sults due to ongoing somatic mutations of immunoglobulingenes22,23. Our PCR assay provided a level of sensitivity that opti-mally distinguished between pre-vaccination PBMC samples,which were uniformly PCR-positive, and samples obtained aftervaccination, which have been continuously PCR-negative in 8 ofthe 11 patients with MBR-positive tumors. Although the long-term clinical importance of molecular remissions in FL patientsremains to be determined10,24, vaccination either further reducedthe tumor burden beyond that already achieved by chemother-apy or led to the redistribution of residual tumor to sites otherthan peripheral blood.

However, the correlation between T-cell and molecular re-sponses is not perfect. For example, although the ability ofPBMCs to recognize (patient 11) and kill (patient 19) Au FL cellsis obvious, their fate after encountering such targets in the localtumor bed in vivo is unknown. Alternatively, the occurrence ofmolecular remission without T-cell responses of a magnitudecomparable to other patients (patient 7) may be attributed toother anti-tumor effector mechanisms not being measured.Furthermore, the important, broader questions of what the pre-cise mechanisms of tumor eradication are (such as type 1 or 2cytokines, or apoptosis) and which immunologic assays corre-late with anti-tumor effects are outside of the scope of thisstudy, and will ultimately require larger numbers of patientstreated with this or other effective cancer vaccines. However,the observation that at least three patients (patients 6, 15 and28) achieved molecular remissions without a detectableantibody response indicates that a humoral response was notrequired.

GM-CSF may be an essential component of this vaccine for-mulation’s potency. Studies in animals have demonstrated theability of paracrine GM-CSF, either by gene delivery to tumorcells8 or as free cytokine with defined tumor antigens7, to gener-ate tumor-specific CD8+ T-cell immunity. GM-CSF probably actsby recruiting antigen-presenting cells, including dendritic cells,which may activate pathways of antigen processing that allowexogenous proteins to be presented by class I molecules25. Ourearlier clinical protocol using the same immunogen adminis-tered without GM-CSF showed humoral without CD8+ T-cell re-sponses3,4,26. Although low levels of cytotoxicity were reported inthat earlier study (10% lysis; ref. 26), in retrospect, the evidencefor substantial T-cell activation was not as robust or convincing,in comparison with our results here.

In conclusion, our analysis of molecular response rate pro-vides definitive evidence for an anti-tumor effect of lymphoma-specific vaccination. Most patients are capable of generating atumor-specific T-cell response capable of clearing residual tumorcells from the blood after tumor cytoreduction by chemother-apy. Our results thus provide the rationale for a prospective ran-domized multicenter trial planned by the National CancerInstitute comparing chemotherapy alone with the samechemotherapy followed by idiotypic protein keyhole limpet he-mocyanin vaccination with GM-CSF, with remission duration asan endpoint.


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m


ARTICLES

collected, and cytokine release was measured by ELISA (R&D Systems,Minneapolis, Minnesota). Cytokine responses were defined as twice ormore of the amount produced by negative controls.

To evaluate specificity and MHC restriction, PBMCs and autologous nor-mal B cells, isolated from PBMCs by FACS, using a flouoresceinated mono-clonal antibody against the light chain not expressed by the patient’s FLcells, were cultured in a similar way. Autologous normal B cells were iso-lated with greater than 95% purity and viability, and were capable of takingup radioisotope with minimal spontaneous release. FL cells were pre-treated for 16 h with 50 µg/ml monoclonal antibody against MHC class I orII (G46-2.6 and TU39, respectively; Pharmingen, San Diego, California) orcontrol monoclonal antibody (R2B-8 and MARG2c-3, respectively; Sigma)before being co-cultured with PBMCs. In parallel, PBMCs were also culturedwith the same monoclonal antibodies described above and either FL-de-rived immunoglobulin or a panel of control, isotype-matched FL im-munoglobulins. CD4+ and CD8+ T cells (1 × 106 of each) were positivelyselected from immune PBMCs by magnetic cell sorting (MACS; MiltenyiBiotec, Auburn, California), with a purity of more than 95%.

Tumor-specific, direct cytotoxicity. CD40L pre-stimulation of Au FL cellswas done as described29,30. Au FL cells were co-cultured for 3–5 d with ad-herent, irradiated (96 Gy) CD40L-transfected fibroblasts (a gift from J.Schultze, Dana Farber Cancer Institute, Boston, Massachusetts) and thenwere irradiated (64 Gy), and used as stimulators for pre- and post-immunePBMCs in parallel for 5 d. After this single cycle of stimulation, PBMCs wererecovered and used as effectors in a 12-hour indium release assay31.Unmodified Au FL, autologous, normal B cells (see above), or autologouslymphoblastoid line targets were labeled with 111Indium-oxine (370MBq/ml; MPI Pharmacy/Amersham, Arlington Heights, Illinois) and plated(1 × 104 cells/well) with effectors in triplicate. Data are reported as mean %specific lysis + s.e.m. Spontaneous release was less than 25% for FL and30–40% for normal B and lymphoblastoid line targets.

Au FL targets were pre-treated for 2 h with 50 µg/ml of monoclonal anti-bodies against MHC class I or II or control monoclonal antibodies (de-scribed above), washed, and then used in the CTL assay. Effector CD8+ orCD4+ T cells were positively selected from immune PBMCs at the end of thesingle 5-day stimulation by magnetic cell sorting (as described above).

Acknowledgments

We thank the physicians and nursing staff of the former Biological Response

Modifiers Program and the pharmacy and nursing staff of the 13E unit in

building 10, NIH Clinical Center, for their patient care. We also thank S. Grove,

J. Mikovits and C. Petrow for technical assistance and D. D. Taub for scientific

discussions.

RECEIVED 22 APRIL; ACCEPTED 2 AUGUST 1999

1. Stevenson, G.T. & Stevenson, F.K. Antibody to molecularly-defined antigen con-fined to a tumor cell surface. Nature 254, 714–716 (1975).

2. Lynch, R.G. et al. Myeloma proteins as tumor-specific transplantation antigens.Proc. Natl. Acad. Sci. USA 69, 1540–1544 (1972).

3. Kwak, L.W. et al. Induction of immune responses in patients with B-cell lym-phoma against the surface-immunoglobulin idiotype expressed by their tumors.N. Engl. J. Med. 327, 1209–1215 (1992).

4. Hsu, F.J. et al. Tumor-specific idiotype vaccines in the treatment of patients withB-Cell lymphoma–Long term results of a clinical trial. Blood 89, 3129–3135(1997).

5. Greenberg, P.D. Adoptive T-cell therapy of tumors: mechanisms operative in therecognition and elimination of tumor cells. Adv. Immunol. 49, 281–355 (1991).

6. Shu, S., Chou T. & Rosenberg, S.A. Generation from tumor-bearing mice of lym-

phocytes with in vivo therapeutic efficacy. J. Immunol. 139, 295–304 (1987).7. Kwak, L.W., Young, H.A., Pennington, R.W. & Weeks, S.D. Vaccination with syn-

geneic, lymphoma-derived immunoglobulin idiotype combined with granulo-cyte-macrophage colony-stimulating factor primes mice for a protective T-cellresponse. Proc. Natl. Acad. Sci. USA 93, 10972–10977 (1996).

8. Dranoff, G. et al. Vaccination with irradiated tumor cells engineered to secretemurine granulocyte-macrophage colony-stimulating stimulates potent, specific,and long-lasting anti-tumor immunity. Proc. Natl. Acad. Sci. USA 90, 3539–3543(1993).

9. Lee, M.S. et al. Detection of minimal residual disease carrying the t(14;18) byDNA sequence amplification. Science 237, 175–178 (1987).

10. Gribben, J.G. et al. All advanced stage non-Hodgkin’s lymphomas with a poly-merase chain reaction amplifiable breakpoint of bcl-2 have residual cells contain-ing the bcl-2 rearrangement at evaluation and after treatment. Blood 78,3275–3280 (1991).

11. Cortopassi, G.A. & Arnheim, N. Using the polymerase chain reaction to estimatemutation frequencies and rates in human cells. Mutat. Res. 277, 239–249 (1992).

12. Rosenberg, S.A. et al. Immunologic and therapeutic evaluation of a syntheticpeptide vaccine for the treatment of patients with metastatic melanoma. NatureMed. 3, 321–327 (1998).

13. Correale, P. et al. In vitro generation of human cytotoxic T lymphocytes specificfor peptides derived from prostate-specific antigen. J. Natl. Cancer Inst. 89,293–300 (1997).

14. Mandruzzato, S., Brasseur, F., Andry, G., Boon, T. & van der Bruggen, P. A. CASP-8 mutation recognized by cytolytic T lymphocytes on a human head and neckcarcinoma. J. Exp. Med. 186, 785–793 (1997).

15. Kato, K., Cantwell, M.J., Sharma, S. & Kipps, T.J. Gene transfer of CD40-ligand in-duces autologous immune recognition of chronic lymphocytic leukemia B cells. J.Clin. Invest. 101, 1133–1141 (1998).

16. Nair, S.K. et al. Induction of primary carcinoembryonic antigen (CEA)-specific cy-totoxic T lymphocytes in vitro using human dendritic cells transfected with RNA.Nature Biotechnol. 16, 364–369 (1998).

17. Brossart, P. et al. Her-2/neu-derived peptides are tumor-associated antigens ex-pressed by human renal cell and colon carcinoma lines and are recognized by invitro induced specific cytotoxic T lymphocytes. Cancer. Res. 58, 732–736 (1998).

18. Kim, H.T. et al. γδ T cell recognition of tumor Ig peptide. J. Immunol. 154,1614–1623 (1995).

19. Hsu, F.J. et al. Vaccination of patients with B-cell lymphoma using autologousantigen-pulsed dendritic cells. Nature Med. 2, 52–58 (1996).

20. Kwak, L.W. et al. Transfer of myeloma idiotype-specific immunity from an activelyimmunized marrow donor. Lancet 345, 1016–1020 (1995).

21. Osterborg, A. et al. Idiotype immunization combined with granulocyte-macrophage colony-stimulating factor in myeloma patients induced type I, majorhistocompatibility complex-restricted CD8- and CD4-specific T-cell response.Blood 91, 2459–2466 (1998).

22. Raffeld, M., Neckers, L., Longo, D.L. & Cossman, J. Spontaneous alteration of id-iotype in a monoclonal B-cell lymphoma. Escape from detection by an anti-idio-type. N. Engl. J. Med. 312, 1653–1658 (1985).

23. Meeker, T. et al. Emergence of idiotype variants during treatment with anti-idio-type antibodies. N. Engl. J. Med. 312, 1658–1665 (1985).

24. Lopez-Guillermo, A. et al. The clinical significance of molecular response in indo-lent follicular lymphomas. Blood 91, 2955–2960 (1998).

25. Grabbe, S., Beissert, S., Schwarz, T. & Granstein, R.D. Dendritic cells as initiatorsof tumor immune responses: a possible strategy for tumor immunotherapy?Immunol. Today 16, 117–120 (1995).

26. Nelson, E.L. et al. Tumor-specific, cytotoxic T-lymphocyte response after idiotypevaccination for B-cell, non-Hodgkin’s lymphoma. Blood 88, 580–589 (1996).

27. Fisher, R.I. et al. Diffuse aggressive lymphomas: increased survival after alternat-ing flexible sequences of ProMACE and MOPP chemotherapy. Ann. Intern. Med.98, 304–309 (1983).

28. Kobrin, C.B. & Kwak, L.W. Development of vaccine strategies for the treatment ofB-cell malignancies. Cancer Invest. 15, 577–587 (1997).

29. Schultze, J.S. et al. Follicular lymphomas can be induced to present alloantigenefficiently: a conceptual model to improve their tumor immunogenicity. Proc.Natl. Acad. Sci. USA 92, 8200–8204 (1995).

30. Schultze, J.S., Seamon, M.J., Michalak, S., Gribben, J.G. & Nadler, L.M.Autologous tumor infiltrating T cells cytotoxic for follicular lymphoma cells canbe expanded in vitro. Blood 89, 3806–3816 (1997).

31. Wiltrout, R.H., Taramelli, D. & Holden, H.T. Measurement of macrophage-medi-ated cytotoxicity against adherent and non-adherent target cells by release of111Indium-oxine. J. Immunol. Meth. 43, 319–331 (1981).


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://m

edic

ine.

nat

ure

.co

m

Documents

document