Mannan-MUC1^Pulsed Dendritic Cell Immunotherapy: APhase ITrial in Patientswith AdenocarcinomaBruce E. Loveland, Anne Zhao, ShaneWhite, Hui Gan, Kate Hamilton, Pei-Xiang Xing, GeoffreyA. Pietersz,VassoApostolopoulos, HilaryVaughan,Vaios Karanikas, Peter Kyriakou, Ian F.C. McKenzie,and Paul L.R. Mitchell
Abstract Purpose: Tumor antigen-loaded dendritic cells show promise for cancer immunotherapy.This phase I study evaluated immunization with autologous dendritic cells pulsed withmannan-MUC1 fusion protein (MFP) to treat patients with advanced malignancy.Experimental Design: Eligible patients had adenocarcinoma expressing MUC1, were ofperformance status 0 to 1, with no autoimmune disease. Patients underwent leukapheresis togenerate dendritic cells by culture ex vivo with granulocytemacrophage colony-stimulating factorand interleukin 4 for 5 days. Dendritic cells were then pulsed overnight with MFP and harvestedfor reinjection. Patients underwent three cycles of leukapheresis and reinjection at monthly inter-vals. Patients with clinical benefit were able to continuewith dendritic cell-MFP immunotherapy.Results:Ten patients with a range of tumor types were enrolled, with median age of 60 years(range, 33-70 years); eight patients were of performance status 0 and two of performancestatus 1. Dendritic cell-MFP therapy led to strong T-cell IFNg Elispot responses to the vaccineand delayed-type hypersensitivity responses at injection sites in nine patients who completedtreatments. Immune responses were sustained at 1 year in monitored patients. Antibodyresponses were seen in three patients only and were of low titer. Side effects were grade 1 only.Two patients with clearly progressive disease (ovarian and renal carcinoma) at entry were stableafter initial therapy and went on to further leukapheresis and dendritic cell-MFP immunotherapy.These two patients have now each completed over 3 years of treatment.Conclusions: Immunization produced T-cell responses in all patients with evidence of tumorstabilization in 2 of the 10 advanced cancer patients treated. These data support further clinicalevaluation of this dendritic cell-MFP immunotherapy.
Dendritic cells as vectors for antigen delivery are a major focusof cancer immunotherapy. Early clinical studies have useddiverse strategies, including the ex vivo loading of autologousmonocyte-derived dendritic cells with antigens (syntheticpeptides and recombinant proteins in melanoma, prostate,gastrointestinal, lung and breast cancer; refs. 1–6), with tumorcell lysates (in kidney and prostate cancer; refs. 7, 8), with DNA(9) or RNA (10), and with dendritic cell/tumor cell coinjection(11) or fusions (12). Patient benefit in these studies has beeninfrequent and inconsistent (reviewed in ref. 13), indicating
the challenges to be surmounted if cellular therapy is tobecome clinically relevant.
We now report clinical and laboratory outcomes of a phase Iclinical trial using monocyte-derived dendritic cells loaded witha mucin 1 (MUC1) VNTR recombinant protein conjugated tooxidized mannan [mannan-MUC1 fusion protein (MFP)]. TheMUC1 antigen was chosen as it is highly and abnormallyexpressed with aberrant glycosylation by many epithelialtumors including breast, colon, kidney, lung, esophageal,stomach, and ovarian cancer. On normal tissues, low levelexpression of heavily glycosylated MUC1 is restricted to theapical cell surface facing the lumen of ducts and glands (14).Preclinical studies of the recombinant (not glycosylated)MUC1 antigen in mice indicated a particular advantage ofconjugation to oxidized mannan, with enhanced stimulationof cellular immune responses (15). Unique features ofmannan-conjugated antigen include targeting the complex tothe mannose receptor of dendritic cells. Oxidized mannanrapidly traverses the endosome to deliver conjugated proteins(e.g., MUC1 VNTR) to the class I presentation pathway (16).However, in prior phase I clinical trials when MFP was injected(i.m., s.c., or i.p.) into >100 patients with advanced metastaticMUC1+ adenocarcinoma, while no toxicity was noted, therewas a predominance of antibody-mediated, rather than cell-mediated, immunity and no clinical responses were observed
Cancer Therapy: Clinical
Authors’Affiliation:AustinResearch Institute, andMedicalOncologyUnit, AustinHospital, Heidelberg, Melbourne,Victoria, AustraliaReceived 7/22/05; revised11/13/05; accepted11/16/05.Grant support: The Austin Research Institute, AusIndustry START grantGRA02245; PRIMA BioMed Ltd., Kew,Victoria, Australia; National Health andMedical Research Council Senior Research Fellowship (B. Loveland); and NationalHealth and Medical Research Council CJMartin Research Fellowship, Australia(V. Apostolopoulos).The costs of publication of this article were defrayed in part by the payment of pagecharges.This article must therefore be hereby marked advertisement in accordancewith18 U.S.C. Section1734 solely to indicate this fact.Requests for reprints: Bruce Loveland,The Austin Research Institute, StudleyRoad, Heidelberg, Victoria 3084, Australia. Phone: 61-3-9287-0666; Fax: 61-3-9287-0600; E-mail: [email protected].
F2006 American Association for Cancer Research.doi:10.1158/1078-0432.CCR-05-1574
www.aacrjournals.org Clin Cancer Res 2006;12(3) February1, 2006869
(17–19).1 Mice and humans differ markedly in their immuneresponses to MUC1. We had shown that the naturallyoccurring human anti-Gala(1,3)Gal antibodies in cancerpatients cross-reacted with MUC1 peptides (20) and consid-ered it likely that anti-Gal antibodies formed immunecomplexes with injected MFP, bound Fc receptors, andstimulated antibody production rather than cell-mediatedimmunity (21). This was clearly seen in Gal�/� mice (whichin this regard resemble humans) where, in contrast to normal(Gal+/+) mice, the responses were predominantly antibody. Thetrend to antibody production in Gal�/� mice was reversed bythe ex vivo loading of antigen-presenting cells with MFP (21).These studies led to the current clinical trial using autologousdendritic cells as vaccine carriers, aiming to maximize cellularimmunity while minimizing the induction of antibody, whichseemed to have no clinical benefit in the advanced diseasesetting (17–19).This phase I study in patients with MUC1+ adenocarcinoma
has shown that MFP loaded ex vivo into autologous monocyte-derived dendritic cells was without significant side effects,reliably led to T-cell immune responses, and was correlatedwith stabilization of disease in some patients.
Materials andMethods
Study designThe primary objective of the study was to assess toxicity
with secondary objectives of assessing antitumor efficacy, immuneresponses, and procedure feasibility.
Eligible patients underwent leukapheresis at 4-weekly intervals onthree occasions to obtain dendritic cell precursors. Autologousdendritic cells were cultured ex vivo and injected into patients atintradermal and s.c. sites. When it became evident that some patientswere deriving clinical benefit, a protocol amendment allowed furtherdendritic cell-MFP immunotherapy.
PatientsEligibility criteria were adenocarcinoma with no available curative
therapy (i.e., either stage IV or with relapsed disease); no centralnervous system involvement; no chemotherapy, radiotherapy, immu-notherapy, or experimental treatment within the previous 4 weeks; lifeexpectancy of >12 weeks; Eastern Cooperative Oncology Groupperformance status of 0 to 1; no other malignancy within 5 yearsexcept basal cell carcinoma of skin or noninvasive cervical cancer; noconcurrent systemic corticosteroid therapy; adequate hematologic (Hb,>10 g/dL; WBC, >3 � 109/L; platelets, >109/L), renal (creatinine, <0.14mmol/L), and hepatic function (bilirubin, <60 mmol/L), and noautoimmune disease apart from autoimmune thyroid disease. Allpatients were required to have strongly expressed MUC1 (VNTR)antigen on stored fixed tumor specimens. Tumor tissue was stained andscored for MUC1 expression on normal and cancer cells using theimmunoperoxidase technique and four anti-MUC1-VNTR monoclonalantibodies (VA1, VA2, BC2, and 3E1.2, which recognize differentepitopes, as reviewed in ref. 22). Patients gave written informedconsent and the study was approved by the institutional HumanResearch Ethics Committee.
Investigations and monitoringBaseline investigations were history and examination, blood tests
for hematology, and clinical chemistry including electrolytes, renal
and hepatic function, relevant tumor markers, autoantibody screen,
and HLA typing. Hepatitis B and C and HIV serology were required
to be negative for patients to undergo leukapheresis. Baseline
imaging of disease sites was done within 3 weeks of study entry.
Following each injection of cell product, patients were monitored for
3 hours. Patients were reviewed thrice at monthly intervals during
the 12-week study period, including blood tests and repeat imaging
of disease sites at 12 weeks. Patients who received ongoing therapy
were reviewed with repeat imaging 3-monthly and monthly tumor
markers.
Preparation of vaccine and test antigensMFP vaccine was prepared as previously described (23, 24) The
cDNA insert (pDF9.3; ref. 25) encoded 103 amino acids of the humanMUC1 protein, including three copies of the VNTR motif (PDTRPAPG-STAPPAHGVTSA) flanked by natural variants of this sequence.Recombinant MFP, consisting of glutathione-S-transferase (26 kDa)and the VNTR-containing sequences (103 amino acids; 12 kDa), wasproduced in E. coli , solubilized by sonication, and purified by elutionfrom glutathione-agarose columns (Sigma, St. Louis, MO). Mannan(Sigma) was oxidized with sodium periodate and conjugated to fusionprotein to make MFP (23, 24). For in vitro assays of MUC1 vaccine–specific immune responses in delayed-type hypersensitivity (DTH),Elispot, and ELISA assays, an alternative source of recombinant MUC1antigen (pVNTR), not containing glutathione-S-transferase, was pre-pared using the pTrcHisB vector (Invitrogen) and purified on a Nickel-NTA column (Qiagen, Melbourne, Australia). MFP and pVNTR weretested for sterility and for endotoxin content by the LAL assay to be <30endotoxin units/250 Ag protein.
Generation of dendritic cells in ex vivo cell culture,antigen loading, and injectionPatients underwent three leukaphereses at 4-weekly intervals to
generate dendritic cells for three sets of injections within the 12-weektrial period. Peripheral blood mononuclear cells (PBMC) werecollected without priming over 2.5 to 3.5 hours using a HaemoneticsMCS plus machine, were enriched for dendritic cells using standardmonocyte adherence methods (reviewed in ref. 26), and culturedwith recombinant human granulocyte macrophage colony-stimulatingfactor (500 units/mL; Schering-Plough Ltd., Sydney, Australia) andrecombinant human interleukin 4 (500 units/mL; Peprotech, Inc.,Rocky Hill, NJ) in AIM-V medium (Life Technologies, Gaithersburg,MD) and 2% v/v autologous serum. The MFP vaccine (10 Ag/mL)was added to cell cultures on day 5 before recovering nonadherentdendritic cells on day 6, which were released for injection based onculture morphology (>50% dendritic cell-like) and Gram stain(negative). The cells were resuspended to 4 � 107/mL in BP salinefor injection (Astra Pharmaceuticals, Sydney, Australia) with 2% v/vautologous serum, and injected into two intradermal sites (107 viablecells per site) with the remainder divided between two to four s.c.sites in the upper arms and/or legs. A sample was analyzed forphenotype by flow cytometry as described below.
Assessment of immune responsesPBMC and serum samples were cryopreserved at six time points: at
each leukapheresis (weeks 1, 5, and 9) and 3 weeks after each
injection (weeks 4, 8, and 12). Where possible, samples were again
taken at 6 and 12 months. The intradermal injection sites were
assessed at 48 hours for erythema and edema. In five patients (A07-
A11), a protocol amendment allowed two skin biopsies of injection
sites to be taken at week 10, with one formalin fixed for histology
and the other cryopreserved and sectioned for immunohistochemistry
(staining with anti-CD4, anti-CD8, and anti-CD86 monoclonal
antibodies) to assess DTH. In five patients (A01-A05), skin tests
at week 12 were used to assess general cellular immunity using
MCI Multitest Kits (Pasteur Merieux, Lyon, France) for Tetanus,
Trichophyton (mentagrophytes), and Proteus (mirabilis) antigens. In
these patients, recombinant pVNTR (10 Ag injected i.d.) was also
used to assess DTH to the vaccine, recording the dimensions
of erythema and edema for each antigen 24 hours after applying
the test.Elispot cytokine assay of PBMC T cells and test antigens. A standard
Elispot assay of thawed PBMC was used (27). In brief, 96-well plates(MAIPS S45, Millipore, Sydney, Australia) were precoated with 50 AL ofantihuman IFNg antibody (1D1K, Mabtech, Melbourne, Australia) andblocked with 10% v/v human AB serum. Each test well of triplicategroups contained 5 � 105 PBMC in 100 AL RPMI 1640 with 2.5% v/vhuman AB serum (Valley Biomedical, Winchester, VA) plus 25 AL ofantigen to a final concentration of 20 Ag/mL or medium (control).Where cell numbers allowed, CD4+ or CD8+ T cells were depleted fromPBMC using Dynal magnetic beads (depletion >98% assessed by flowcytometry; data not shown) before the Elispot assay. Antigens werepurified protein derivative (Tuberculin, Staten Serums Institute,Copenhagen, Denmark; positive control antigen) and recombinantpVNTR (specific test antigen). The plates were incubated for 18 hours at37jC in 5% CO2, washed with PBS containing 0.1% v/v Tween 20, thenwith PBS, and incubated with biotin-conjugated antihuman IFNg
antibody (7-B6-1, Mabtech) for 2 hours at room temperature beforewashing. Bound cytokine was visualized using 1 Ag/mL streptavidin-alkaline phosphatase conjugate (Mabtech) incubated for 2 hours atroom temperature, then incubated with the colorimetric AP detection kitaccording to the instructions of the manufacturer (Bio-Rad, Hercules,CA). Cytokine spots (spot-forming units) were counted with the AIDElispot Reader system (Autoimmun Diagnostika GmbH, Strassberg,Germany).
Serum antibody assays. Anti MUC1 VNTR antibodies in patient serawere measured by ELISA against pVNTR antigen as per the publishedmethod (19). A positive control serum (IFCM10_2EC) was selectedfrom a panel of patients previously immunized with MFP who haddeveloped anti-VNTR antibodies. A negative control serum (ARI_S2P)was selected, being from a normal individual with no detectable anti-VNTR antibody.
Flow cytometric cell analyses. Samples of the injected dendritic cellpopulation and of leukapheresed PBMC, in aliquots of 3 � 105, wereanalyzed by flow cytometry (FacsCalibur II, Becton Dickinson, SanJose, CA) for phenotypic markers, in particular to assess dendriticcells content and maturity, using fluorochrome-conjugated monoclo-nal antibodies according to the recommendations of the manufac-turer [anti-CD3.FITC, CD14.TC, CD40.FITC, CD80.TC, CD83.APC,CD86.PE, MHC-II(DR).APC (Caltag Laboratories, Burlingame, CA)],and anti-CD83.FITC (Becton Dickinson)] or using monoclonalantibody supernatants [W6/32 anti-MHC-I (ARI stocks), CMRF-44(28), kindly provided by D. Hart, Master Medical Research Institute,Brisbane, Australia, and anti-CCR7, kindly provided by F. Sallusto,Institute for Research in Biomedicine, Bellinzona, Switzerland] withFITC-conjugated antimouse immunoglobulin. The mannose receptorwas detected using FITC-conjugated mannan (29).
Results
Patients. The characteristics of the 10 patients entered onstudy are detailed in Table 1. Median age was 60 years (range,33-70 years) and performance status was 0 in eight patientsand 1 in two patients. Tumor types (MUC1+) were renalcarcinoma in three patients, breast carcinoma in two, ovarian/fallopian tube in two, and one patient each with non–small-cell lung, colon, and esophageal carcinoma. Eight patients hadstage IV disease whereas two patients (A03 and A09) hadrecurrent, incurable stage III disease. All patients had receivedprior therapy in addition to surgery, which included chemo-
therapy in five patients, radiotherapy in four, IFN in two, andhormonal therapy in one patient.
Treatment. Nine of 10 patients completed all three roundsof leukapheresis, cell culture, and immunization over the 12-week study period, with the third leukapheresis of A01 aband-oned due to poor vascular access. From these 29 leukaphereses,the mean PBMC yield was 2.8 F 1.4 (SE) � 109 cells and after 6days of culture with granulocyte macrophage colony-stimulat-ing factor and interleukin 4, the mean viable cell recovery(‘‘dendritic cells’’) was 8.8 F 4.9 � 107 cells. However, withinthis broad range of cell recoveries, there were reproducible‘‘patient-specific’’ features; i.e., some individuals had consistentlow PBMC yields but high dendritic cells recoveries (e.g., A11:1.9 F 0.4 � 109 PBMC; and 15.3 F 3.4 � 107 dendritic cells,being 7.9% of the starting number) or the converse (e.g., A09:4.9 F 0.6 � 109 PBMC; and 5.8 F 1.3 � 107 dendritic cells,being 1.2% of the starting number). The content of dendriticcells in the harvested cultures varied considerably (estimated tobe 60-85% based on flow cytometry as CD86+CD14�CD3� asdetailed below) and the residual nondendritic cells (largelyT cells by CD3 staining) were not expected to affect theimmunogenicity of the injected cells. All recovered MFP-loadedcultured cells were injected into the patients: 10 million cellsinto each of two intradermal sites in the upper arm, with theremainder divided equally between two to four (for higher cellrecoveries) s.c. sites.
Toxicity. The toxicity of leukapheresis and cell injectionswas mild. Two patients had anxiety and lower back pain relatedto leukapheresis. There was no clinical toxicity of the vaccine.One patient developed elevation of antithyroid antibodies upto a maximal titer of 1:400 with no evidence of clinicalautoimmune disease, and another had anti–smooth muscleantibodies at 1:80.
Antitumor efficacy. In four patients (A02, A08, A09, andA10), disease remained stable during the 12-week study period(Table 1). This included two patients, with breast and coloncarcinoma, respectively, who had stable or slowly progressingdisease at study entry, which was not considered to have beenaltered by study therapy. The other two patients, with ovarian(A09) and renal carcinoma (A08), had documented progressivedisease at study entry and their clinical courses are presentedhere in detail.
Patient A09, a 64-year-old woman, presented with stage IIIovarian carcinoma 26 months before study entry, andunderwent debulking surgery followed by six cycles ofcarboplatin and paclitaxel chemotherapy. Incurable recurrentdisease, diagnosed by elevated serum CA125, occurred 20months from presentation and she was treated with fourcycles of carboplatin, with normalization of the CA125 level.Study therapy commenced 3 months following completion ofchemotherapy, at which time serum CA125 remained normalat 14 (normal <35) with no evaluable disease on computedtomography scanning (as typical of many ovarian cancercases). After 6 weeks on study, serum CA125 rose to 47 IU/mL (normal <35), then to 64 at 2 months from study entry,indicating definite progressive disease during the early stageof trial therapy. Following the second dendritic cell-MFPinjection, CA125 stabilized and remained in the range of 57to 70 for 4 months, after which the level again began toincrease, increasing from 64 to 165 over 4 months (indicatedas periods ii and iii in Fig. 1 using extrapolated trendlines).
MUC1-Pulsed Dendritic Cell Cancer Immunotherapy
www.aacrjournals.org Clin Cancer Res 2006;12(3) February1, 2006871
In view of the observed 4-month stabilization, thenincreasing CA125 levels, approval for a protocol amendmentwas obtained from the Institutional Human Research EthicsCommittee to allow for further therapy. This resulted in adelay of 7 months before restarting intradermal dendritic cell-MFP immunotherapy, first at monthly intervals (threeinjections), then 3-monthly (using cryopreserved MFP-loadeddendritic cells). Serum CA125 on recommencing studytherapy was 165, increasing to 191 during the next month.After the second of the additional injections at 11 months,the patient had largely stable disease for a period of 18months, although with very slowly increasing CA125 levelswhich remained in the range of 182 to 258 IU/mL (Fig. 1,period iv). From f27 months, there has been an inexorableincrease in CA125 (most recently to 1,256 IU/mL, 34 monthsfrom recommencing therapy and 43 months since studyentry), interspersed with brief intervals where dendritic cell-
MFP injections are associated with a transient reduction inserum CA125 (e.g., at 31, 35, and 40 months; Fig. 1).Reversion to monthly injections at 35 months seemed toslow the CA125 increase for 5 months only. Through >3years, there has been no evaluable disease on computedtomography scanning.The second patient, A08, a 69-year-old man, underwent right
nephrectomy for a clear cell carcinoma of the kidney 22 monthsbefore study entry. The 7-cm tumor was adherent to the liverand extended into perinephric fat. The patient receivedadjuvant IFN for 6 months but stage IV disease with metastasesto mediastinal nodes and liver was noted at 14 months fromdiagnosis. Computed tomography scanning at study entryshowed progressive disease in mediastinal nodes (Fig. 2A, C,and E). The disease sites remained stable through the 12-weekinitial study period. After an interval of 8 months, during whichthe patient received no therapy and disease remained stable, thepatient recommenced leukapheresis with three dendritic cell-MFP injections at monthly intervals followed by ongoing3-monthly injections. Computed tomography scanning at 14months showed the earlier disease sites to be stable or reducedin size (Fig. 2B, D, and F). The patient remained clinically wellon dendritic cell-MFP therapy, with overall stable disease 36months from commencing study therapy. At 20 months, anewly enlarged 2.0-cm aorto-pulmonary window lymph nodewas noted, which later increased to 3.5 cm in diameter butresolved by 36 months.Dendritic cell yield and phenotype. In preliminary culture ex-
periments to establish procedural details, the numbers and pro-portionofmonocyte-derived dendritic cells (CD86+CD14�CD3�)increased markedly from day 4 to day 6 but little more today 8 (data not shown). Because the cell yield wasconsidered important, a 6-day culture period was used.Addition of the MFP antigen was on day 5, with rapiduptake into dendritic cells as monitored by surface andintracellular staining with anti-VNTR monoclonal antibodiesafter 0.5, 2, 6, and 24 hours (data not shown). Although nospecific steps to mature the dendritic cells were undertaken,CD40, CD83, CD86, and CMRF44 expression on MFP-loadeddendritic cells tended to be greater than on untreated
Table1. Patient characteristics, immunologic responses, and antitumor efficacy
ID Age/sex HLA Primary adenocarcinoma diagnosis Best response DTH* Elispot ELISAc
*DTH reactions at the dendritic cells intradermal injection sites are indicated following the specified injection times.cLow titer (1/100) anti-VNTR serum antibody reactions or none detectable, measured within 6 months of treatment.bStable or slow progression at study entry and hence not considered to have clinical benefit from dendritic cell-MFP therapy.x Received ongoing dendritic cell-MFP treatment. Overall slow progression (A09) or stable with transient episode of resolving progression (A08).
Fig. 1. Serum CA125 levels in patient A09, monitored from pre-trial to 42 months,with the 23MFP/dendritic cell injections (arrows) below the X axis. Injections weregiven as two triple injection series every monthwith a 6 month interval, then every3 months to 33 months, thenmonthly.The dashed linear trendlines (MicrosoftExcel), for periods of relatively systematic CA125 levels, indicate marked reductionsin the rate of CA125 increase following the early injection series (sections ii and iv,marked by vertical lines) and acceleration from 28 months (section v).
Cancer Therapy: Clinical
www.aacrjournals.orgClin Cancer Res 2006;12(3) February1, 2006 872
dendritic cells (Fig. 3) whereas other phenotypic markers didnot change (data not shown).Dendritic cell– like cells were largely CD3�CD86+ and
contained minimal CD3+ and CD14+, high expression ofMHC class I and class II, and high CD86, whereas CD83expression was low to moderate and CMRF44 expression wasmoderate to strong. The mannose receptor, detected by thebinding of FITC-conjugated mannan, was usually expressed ona majority of, but not all, dendritic cell– like cells (data notshown). There were no notably variant cultures generated fromthe 29 leukaphereses of these 10 patients.Multitest CMI Kit and anti-VNTR DTH responses. Five
patients were tested for skin DTH responses to seven commonantigens using a Multitest CMI kit and to a local i.d. injectionof the MUC1-VNTR vaccine antigen. All five had skin DTHreaction responses to tetanus and four had weak DTHresponses to the low challenge dose of the VNTR peptide,with variable responses to the other antigens (data not shown).These results showed the participants to be immunocompetentand the DTH responses to MUC1-VNTR injection correlatedwith the in vitro T-cell assays.Elispot assays of IFNg secretion. The ex vivo 18-hour Elispot
assay did not require cell expansion as it detected memoryeffector cells (both CD4+ and CD8+ cytokine-producing). It isa sensitive method that has been used to monitor cellularimmune responses in patients receiving immunotherapy(30, 31). Purified protein derivative was used as a positivecontrol antigen to validate the stored PBMC samples, and towhich all patients were expected to respond.IFNg-secreting T cells were generated promptly in all 10
patients immunized with dendritic cells pulsed with oxidizedmannan-MUC1, were specific for MUC1 pVNTR antigen(with no glutathione-S-transferase), and detected as early as
2 weeks after the first injection (Fig. 4A). The relativeresponses to the purified protein derivative and pVNTRantigens by a cell sample, rather than absolute values, wereconsidered the most appropriate measure of immuneresponse because of interassay variation. The patientsdeveloped specific VNTR IFNg-secreting T cells either 2 weeksafter the first injection (A02, A04, and A11), 4 weeks after thefirst injection (A01, A03, A05, A07, A08, and A09), or 4weeks after the second injection (A10). Follow-up tests after6 to 12 months for A02 to A11 as available, and 2 to 3 yearsfor A08 and A09 during extended treatments, showed a long-term sustained immune response to the vaccine antigen(Fig. 4A). In addition, when PBMC from patients A04 to A11were depleted of CD4+ or CD8+ T cells and analyzed forVNTR-induced IFNg secretion, it was clear that both CD4+
and CD8+ T cells specific for MUC1 VNTR were present in allpatents (A04 and A06 shown in Fig. 4B). Concurrent assaysof intracellular IFNg production by flow cytometry showedthat the responding cells were either CD4+ or CD8+ T cellsand not other cells, excluding a significant natural killer cellresponse (data not shown).
Thus, all patients made vaccine-specific IFNg responseswithin the 12-week trial period after the three dendritic cell-MFP injections and these responses, in the nine patientswho were further monitored, were shown to be sustainedlong-term.
Measurement of antibodies by ELISA assays. Low titer (1/40)immunoglobulin M (IgM) anti-VNTR serum antibodies wereinduced in 3 of 10 participants (A04, A08, and A10) andseroconverted from the second injection to low titer IgG in twoof the patients (data not shown). A09 had unchanging low titreIgG and no IgM throughout the treatment period. There was nosignificant increase in antibody titers when assayed 6 to 12months after the immunizations.
DTH responses. In previous phase I trials of MFP injection(see Karanikas et al. refs. 17, 19), we had assiduously measuredDTH responses: most were weak and the biopsies showed fewinfiltrating cells. In this study, however, all but one patientdeveloped DTH-type responses of 1 to 2 cm at intradermalinjection sites 24 to 48 hours after the second or third injectionof cell product. Patients A08 and A09, who then receivedongoing dendritic cell-MFP therapy, continued to show DTHresponses at injection sites. The exception was patient A01 whohad no DTH after the second injection (although an in vitrocellular immune response; Fig. 4A) and had no further therapydue both to poor venous access precluding apheresis andto rapidly progressive disease. Typically, there was intenseerythema up to 2 cm in diameter, with or without edema.These responses lasted 48 to 72 hours. Biopsy at 48 hours in 4patients showed an infiltration of CD4+ T cells in substantialnumbers whereas rarely could CD8+ cells be detected (data notshown).
Discussion
Targeting dendritic cells to load a mannan-conjugatedrecombinant MUC1 VNTR (MFP) antigen for immunotherapywas well tolerated and was associated with prolongedessentially stable disease in 2 of 10 patients treated who hadadvanced ovarian and renal carcinoma, respectively. Thepatient with ovarian carcinoma was in second relapse at the
Fig. 2. CT thorax scans of patient A08 at trial entry and after14months. Immediatepre-trial CT (A, B, and C) and14 months later (correspondingD, E, and F), afterreceiving six dendritic cell/MFP injections at1, 2, and 3months (initial protocol), and12,13, and14 months (extended treatment).
MUC1-Pulsed Dendritic Cell Cancer Immunotherapy
www.aacrjournals.org Clin Cancer Res 2006;12(3) February1, 2006873
time of study entry, indicating tumor escape from the effectsof chemotherapy ceased 3 months earlier. The patient withrenal cancer had cancer metastatic to liver and lymph nodes.These two patients, with progressive disease at study entry,continued on study therapy for over 3 years.The trial was intended to establish reliable and practical
procedures to generate and use MFP-pulsed autologousmonocyte-derived dendritic cells, to assess vaccine toxicity,
and, by immunologic monitoring, to provide a measure ofvaccine immunogenicity. The culture process minimized theuse of components that are subject to batch variations andregulatory controls (e.g., minimal reagents to enrich monocytesor deplete other cells and no additional cytokines or chemicalsto mature the dendritic cells) and was based on standardmethods to generate dendritic cells from adherent monocyteswith granulocyte macrophage colony-stimulating factor, inter-leukin 4, and autologous serum.This immunotherapy produced cellular responses to the
MUC1 immunogen in all patients treated, a result which hadnot previously been observed after direct injection of theMFP antigen (17–19), emphasizing the important adjuvanteffect of the autologous dendritic cells to stimulate a muchstronger immune response. Peripheral blood T cells from all 10patients made IFNg in Elispot assays (frequency f1/1,000)after only two or three dendritic cell-MFP injections at 4-weekintervals. There was an extensive cellular infiltrate as a DTHresponse at the sites of boosting injections and little productionof anti-MUC1 serum antibody. The maintenance of the DTHresponses was consistent with the detection of IFNg Elispotresults in vitro. We suspect that the rapid development of DTHresponses, well established within 24 hours after the third orsubsequent boosting injections, was a result of antigen-loadeddendritic cells being injected and no additional time wasrequired for antigen to be taken up locally, processed, andpresented (which usually takes several hours).The low frequency of a serologic response was consistent
with the ex vivo MFP vaccine uptake and internalization intodendritic cells with transient MUC1 detectable on the cellsurface. Thus, it was likely that only low quantities of MUC1protein were available to induce anti-vaccine serum antibodieswhereas internalized protein was available for processing andpresentation to stimulate T-cell responses.Dendritic cells as antigen carriers are now receiving consid-
erable attention, particularly in tumor immunotherapy where ithas been difficult to obtain either appropriate immuneresponses or to produce tumor regression (13, 32). Apart fromdendritic cells being the optimal antigen-presenting cell, thereare additional advantages in targeting them ex vivo: removingthem from an immunosuppressive cytokine milieu (e.g.,interleukin 10) which does not encourage immune responses;ensuring the target antigen has ample opportunity to be takenup and processed by the dendritic cells; and, after injectionintradermally, providing an opportunity for effector T-cellrecruitment and activation to occur in draining lymph nodesdistant from sites of metastatic disease. Although leukapheresisand cell culture in a GMP facility have logistical difficulties,there was no significant clinical toxicity and the quality of theimmune responses generated, along with the clinical effects,justifies the additional procedures. In this phase I trial, eachpatient received all cells generated (though i.d. doses werestandardized to 20 million cells), which did not allowassessment of cell dosage effects. A current follow-up phase IItrial of this dendritic cell/MFP therapy is using standardizeddoses of 40 � 106 cells per treatment.The T-cell responses to VNTR test antigen were striking, with
all patients developing vaccine-specific IFNg-secreting T cells,both CD4+ and CD8+, which was not unexpected. Theparticular recombinant VNTR antigen contains 103 aminoacids from MUC1 protein (three copies of the conserved
Fig. 3. Phenotypic changes tomonocyte-deriveddendritic cells followingpulsingwithMFP: representative results fromtwopatients(A10andA11).Dendriticcellswerecultured for 5 days in vitro and then pulsed for an additional18 hours with10 Ag/mLMFP ( ) or without MFP ( ). Harvested cells were stained with the panel ofantibodies and flow cytometry data for CD40, CD83, CD86, and CMRF44 areshownhere. Large mononuclear cells as determined by thewide and forward anglelight scatter profile. Appropriate isotype controls (matching the isotype of thespecific antibody) were included, with MFP ( ) and noMFP (- - - -).
Cancer Therapy: Clinical
www.aacrjournals.orgClin Cancer Res 2006;12(3) February1, 2006 874
20-mer VNTR motif plus two flanking variants), providingconsiderable epitope diversity. Thus, both CD4 ‘‘helper’’and CD8 ‘‘cytotoxic’’ responder cells were generated in theHLA-diverse patient group (Table 1), as proven by Elispot
(IFNg) and Intracellular Cytokine assays of cell subsets. Twoquestions arise: why were the dendritic cell-MFP so effective instimulating responses, and why was the response not eradicat-ing tumors? Mannan-antigen targeted dendritic cells are likely
Fig. 4. Stimulation of serial PBMC specimensto assess vaccine immunogenicity, measuredasT-cell secretion of IFNg in an Elispot assay.A , data from all patients, grouped according tomagnitude of responses [spot-forming units(SFU)]. Patients received injections at weeks2, 6, and10, with specimens tested at variousintervals after injections (some at 6 and 9months). Patients A08 and A09 receivedextended treatments and tests to 28 and 38months respectively, as shown. B, tests ofPBMC depleted of CD4 or CD8 cells in 9- or11-week samples from patients A04 and A07.Three stimulation conditions are shown:purified protein derivative antigen (positivecontrol, ),VNTR recombinant protein antigen(n), andno antigen (negative control,5) using5� 105 thawed PBMC per well. w, week ofblood sample; M, months after first injection.
MUC1-Pulsed Dendritic Cell Cancer Immunotherapy
www.aacrjournals.org Clin Cancer Res 2006;12(3) February1, 2006875
to be potent because the mannose receptor enables rapidantigen uptake and its cross-linking initiates some of the prog-ressive dendritic cell maturation processes (Fig. 3),2 which arelikely to provide a direct immunogenic benefit after injection.As to why clinical efficacy was less striking than T-cell
responses, there are a number of potential factors wherebydendritic cell-MFP therapy was not delivered to the rightpatients or in an optimal fashion. First, it is increasingly beingrecognized that the mere detection of T-cell responsesgenerated by an effective immmunization regimen, as usedhere, will not necessarily correlate with in vivo effectorfunction (33) and we, like others, are seeking to identifyrelevant effector T-cell subpopulations and their specificitiesby careful stimulation and epitope mapping studies. Second,all patients had advanced disease and the physiologic effectsof tumor burden may have been excessive (34). In the twopatients with ongoing therapy, the 3-monthly intervalbetween injections may have been too long to achieve agreater antitumor effect. The rate of CA125 change is anaccepted surrogate marker of ovarian cancer progression (35),and initially in patient A09 there seemed to be a goodcorrelation between treatment episodes and response; howev-er, this deteriorated in the 3rd year of treatment even afterreverting to monthly injections (Fig. 1). It is possible thateffector T cells which are (newly) recruited by dendritic cell-MFP injection have only a limited period to destroy targetedcancer cells, and might not have the potential to renew,perhaps affected by the immunosuppressive environment ofthe tumor (36). Thus, protocols where the frequency ofdendritic cell-based immunotherapy injections can be tailoredto a patient with early stage (minimal residual) disease byobservation of clinical response may have the best opportu-nity of a benefit. We note that in patient A09, recent dendriticcell-MFP injections were associated with a drop in CA125levels 1 to 3 months later, which then rebounded, although
the in vitro Elispot assay of unselected PBMC showed nomeaningful change in the T-cell response to the vaccineantigen. Screening for the presence of MUC1+ was based onarchival tissue specimens, rather than specimens of ‘‘current’’tumor. Therefore, several of the trial participants may havedeveloped MUC1-loss mutants.Whereas dendritic cells used in a number of other studies
induced good T-cell responses, the use here of oxidizedmannan-coupled antigen induced high levels and sustainedresponses in all patients that have not been regularlyreported in other studies (reviewed in refs. 37, 38). Theoxidized mannan therefore offers an advantage, likely toresult from targeting the mannose receptor and passage ofthe antigen into proteolytic pathways and class I and class IIpresentation to CD8 and CD4 cells, respectively. Both celltypes are recruited following immunization. The roles ofaldehydes and Schiff bases, created by oxidation of themannan, also seem to be important to the immunogenicityof the MFP (29). Finally, the recombinant MUC1 antigenused, consisting of 103 amino acids, is a relatively complexantigen and clearly was processed into peptides recognizedby T cells from all patients with a broad spectrum of HLAhaplotypes (see Table 1). Therefore, the strategy to use MFP/autologous dendritic cells injections has been shown to bepractical, safe, effective in generating strong cellular immu-nity, and in some patients has given indications of atherapeutic effect.
Acknowledgments
We thank the followingpeople for their valuable helpwith different aspects of theclinical trial: Belinda Bardsley, Sharon France,VanessaWaddell, and Dr. John Bates(PRIMA BioMed Ltd.); Dr. Jayesh Desai andWayne Saunders (Austin Hospital);Lesley Barber, Dr. DominicWall, and Kerrie Stokes (Centre for Blood CellTherapies,Peter MacCallum Cancer Centre); Carla Osinski, HarryAletras, Dr. LiWenjun, andVioleta Bogdanoska (The Austin Research Institute); Dr. Carole Smith,TamlaTait,and staff of theApheresis Service (Austin Hospital); and Louise Keelan and SharonWilson (Department of Histopathology, Austin Hospital). BrendanToohey preparedthe pTRC-VNTR antigen. Thanks also to Dr. Pat Mottram for reviewing themanuscript.
References1. Thurner B, Haendle I, Roder C, et al.Vaccination withmage-3A1peptide-pulsed mature, monocyte-deriveddendritic cells expands specific cytotoxicTcells andinduces regression of some metastases in advancedstage IV melanoma. JExpMed1999;190:1669^78.
2. Lodge PA, Jones LA, Bader RA, et al. Dendritic cell-based immunotherapy of prostate cancer: immunemonitoring of a phase II clinical trial. Cancer Res 2000;60:829^33.
3. Panelli MC,Wunderlich J, Jeffries J, et al. Phase 1study in patients with metastatic melanoma of immu-nization with dendritic cells presenting epitopesderived from the melanoma-associated antigensMART-1and gp100. J Immunother 2000;23:487^98.
4. Hersey P, Menzies SW, Halliday GM, et al. Phase I/IIstudy of treatment with dendritic cell vaccines inpatients with disseminated melanoma. Cancer Immu-nol Immunother 2004;53:125^34.
5. Chen Q, Jackson H, Parente P, et al. Immunodomi-nant CD4+ responses identified in a patient vaccinat-ed with full-length NY-ESO-1 formulated withISCOMATRIX adjuvant. Proc Natl Acad Sci U S A2004;101:9363^8.
6. Maraskovsky E, Sjolander S, Drane DP, et al. NY-ESO-1protein formulated in ISCOMATRIX adjuvant is
a potent anticancer vaccine inducing both humoraland CD8+ T-cell-mediated immunity and protectionagainst NY-ESO-1+ tumors. Clin Cancer Res 2004;10:2879^90.
7. Nestle FO, Alijagic S, Gilliet M, et al.Vaccination ofmelanoma patients with peptide- or tumor lysate-pulsed dendritic cells. NatMed1998;4:328^32.
8. Rieser C, Ramoner R, Holtl L, et al. Mature dendriticcells induce T-helper type-1-dominant immuneresponses in patients with metastatic renal cell carci-noma. Urol Int1999;63:151^9.
9. Rosenberg SA, Zhai Y, Yang JC, et al. Immunizingpatients withmetastatic melanoma using recombinantadenoviruses encoding MART-1or gp100 melanomaantigens. JNatl Cancer Inst1998;90:1894^900.
10.MorseMA,Nair SK,MoscaPJ, et al. Immunotherapywith autologous, human dendritic cells transfectedwith carcinoembryonic antigen mRNA. Cancer Invest2003;21:341^9.
11.O’RourkeMG, JohnsonM, Lanagan C, et al. Durablecomplete clinical responses in a phase I/II trial using anautologous melanoma cell/dendritic cell vaccine.Can-cer Immunol Immunother 2003;52:387^95.
12. Homma S, Matai K, Irie M, et al. Immunotherapyusing fusions of autologous dendritic cells and tumor
cells showed effective clinical response in a patientwith advanced gastric carcinoma. J Gastroenterol2003;38:989^94.
13.Mocellin S. Cancer vaccines: the challenge ofdevel-oping an ideal tumor killing system. Front Biosci2005;10:2285^305.
14. Hollingsworth MA, Swanson BJ. Mucins in cancer:protection and control of the cell surface. Nat RevCancer 2004;4:45^60.
15. ApostolopoulosV, Pietersz GA, Loveland BE, et al.Oxidative/reductive conjugation of mannan to antigenselects forT1orT2 immune responses. Proc Natl AcadSci US A1995;92:10128^32.
16. Apostolopoulos V, Pietersz GA, Gordon S, et al.Aldehyde-mannan antigen complexes target the MHCclass I antigen-presentation pathway. Eur J Immunol2000;30:1714^23.
17. KaranikasV, Hwang LA, Pearson J, et al. AntibodyandTcell responses of patients with adenocarcinomaimmunized with mannan-MUC1fusion protein. J ClinInvest1997;100:2783^92.
18. KaranikasV, LoddingJ,MainoVC,McKenzie IF.Flowcytometric measurement of intracellular cytokinesdetects immune responses in MUC1immunotherapy.Clin Cancer Res 2000;6:829^37.
2 B.E. Loveland, A. Zhao, and V. Karanikas, manuscript in preparation.
Cancer Therapy: Clinical
www.aacrjournals.orgClin Cancer Res 2006;12(3) February1, 2006 876
19. Karanikas V, Thynne G, Mitchell P, et al. MannanMucin-1 Peptide Immunization: Influence of Cyclo-phosphamide and the Route of Injection. J Immun-other 2001;24:172^83.
21. Apostolopoulos V, Osinski C, McKenzie IF. MUC1cross-reactive Gala(1,3)Gal antibodies in humansswitch immune responses from cellular to humoral.Nat Med1998;4:315^20.
22.XingPX,ApostolopoulosV, Pietersz G,McKenzie IF.Anti-mucin monoclonal antibodies. Front Biosci2001;6:D1284^95.
23. Apostolopoulos V, Xing P, Trapani J, McKenzie I.Production of anti-breast cancer monoclonal antibod-ies using a glutathione-S-transferase-MUC1bacterialfusion protein. BrJCancer 1993;67:713^20.
24. Apostolopoulos V, Pietersz G, McKenzie I. Cell-mediated immune responses to MUC1 fusion proteincoupled to mannan. Vaccine 1996;14:930^8.
25. Siddiqui J, Abe M, Hayes D, et al. Isolation andsequencing of a cDNA coding for the human DF3breast carcinoma-associated antigen. Proc Natl AcadSci U S A1988;85:2320^3.
26. Banchereau J, Briere F, Caux C, et al. Immunobiol-ogy of dendritic cells. Annu Rev Immunol 2000;18:767^811.
27. Clay TM, Hobeika AC, Mosca PJ, et al. Assays formonitoring cellular immune responses to activeimmunotherapy of cancer. Clin Cancer Res 2001;7:1127^35.
28. Hock BD, Starling GC, Daniel PB, Hart DN. Char-acterization of CMRF-44, a novel monoclonal anti-body to an activation antigen expressed by theallostimulatory cells within peripheral blood, includ-ing dendritic cells. Immunology 1994;83:573^81.
29. Apostolopoulos V, Barnes N, Pietersz GA,McKenzie IF. Ex vivo targeting of the macrophagemannose receptor generates anti-tumor CTLresponses. Vaccine 2000;18:3174^84.
30. Schmittel A, Keilholz U,Thiel E, Scheibenbogen C.Quantification of tumor-specificT lymphocytes withthe ELISPOTassay. J Immunother 2000;23:289^95.
31. Speiser DE, Pittet MJ, Guillaume P, et al. Ex vivoanalysis of human antigen-specific CD8+ T-cellresponses: quality assessment of fluorescent HLA-A2 multimer and interferon-g ELISPOTassays for pa-tient immune monitoring. J Immunother 2004;27:298^308.
32. Schuler G, Schuler-Thurner B, Steinman RM. Theuse of dendritic cells in cancer immunotherapy. CurrOpin Immunol 2003;15:138^47.
33. Rosenberg SA, Sherry RM, Morton KE, et al.Tumorprogression can occur despite the induction of veryhigh levels of self/tumor antigen-specific CD8+ T cellsin patients with Melanoma. J Immunol 2005;175:6169^76.
34. Dunn GP, Old LJ, Schreiber RD. The three Es ofcancer immunoediting. Annu Rev Immunol 2004;22:329^60.
35. Rustin GJ, Bast RC, Jr., Kelloff GJ, et al. Use ofCA-125 in clinical trial evaluation of new therapeuticdrugs for ovarian cancer. Clin Cancer Res 2004;10:3919^26.
36. Mukherjee P, Ginardi AR, Madsen CS, et al.MUC1-specific CTLs are non-functional within apancreatic tumor microenvironment. Glycoconj J2001;18:931^42.
37. Ridgway D. The first 1000 dendritic cell vaccinees.Cancer Invest 2003;21:873^86.
38. Cranmer LD, Trevor KT, Hersh EM. Clinical appli-cations of dendritic cell vaccination in the treatmentof cancer. Cancer Immunol Immunother 2004;53:275^306.
MUC1-Pulsed Dendritic Cell Cancer Immunotherapy
www.aacrjournals.org Clin Cancer Res 2006;12(3) February1, 2006877
