Ind Bcr/Ablimatinibaintainedsuredt at lowrelapseatinibnts of
ncentra-
Gambacorti-Passeriniet al. Blood Cells, Molecules, and Diseases (2002)28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
Differences between in Vivo and in Vitro Sensitivity to Imatinibof Bcr/Abl� Cells Obtained from Leukemic PatientsSubmitted 05/03/02; revised 05/07/02(Communicated by E. Beutler, M.D., 05/07/02)
Carlo B. Gambacorti-Passerini,1,2 Francesca Rossi,1 Magda Verga,2 Holger Ruchatz,1
Rosalind Gunby,1 Roberta Frapolli,3 Massimo Zucchetti,3 Leonardo Scapozza,4
Silvia Bungaro,1,2 Lucia Tornaghi,1,2 Fabio Rossi,2 Pietro Pioltelli,2
Enrico Pogliani,2 Maurizio D’Incalci,3 and Gianmarco Corneo2
ABSTRACT: Imatinib mesylate (imatinib) inhibits Bcr/Abl, an oncogenic fusion protein. Thein vitro effects oimatinib on BCR/ABL� leukemic cells include inhibition of Bcr/Abl tyrosine phosphorylation, block of preration, and induction of apoptosis. Thein vivo effects of imatinib were evaluated in 12 CML (chronic myeleukemia) patients in blast crisis or accelerated phase who were treated with imatinib. Treatment causedin spontaneous proliferation of leukemic cells in 10 of 12 evaluable patients and the development of apo9 of 11 cases. Imatinib also caused an inhibition of Bcr/Abl autophosphorylation; however, the degree of inobtainedin vivo was substantially lower than that achievedin vitro with similar concentrations of imatinib.seven patients cells could be evaluated at relapse: spontaneous proliferation was no longer inhibited anphosphorylation was comparable or superior to that present at the beginning of treatment, beforeadministration. Plasma imatinib concentrations were not reduced. Leukemic cells obtained at relapse min vitro sensitivity (Bcr/Abl autophosphorylation and proliferation inhibition) to imatinib concentration meain vivo (3 �M or higher), although a partial resistance to the antiproliferative effects of imatinib was presen(0.01–0.3�M) concentrations. In four patients, addition of erythromycin to blood samples obtained atrestored imatinib sensitivity in terms of phosphorylation inhibition, indicating that the majority of plasma imwas not available to cells and probably bound to�1 acid glycoprotein. These data suggest that measuremeBcr/Abl kinase activity in peripheral blood samples may represent a more reliable indicator of active co
ntsvet of
a
/ereds-n-
Vene,
INTRODUCTION
Imatinib mesylate (imatinib) (Glivec) reprsents the first rationally designed drug in the fiof oncology which has been developed to inhan oncogenic molecule: the Bcr/Abl fusion ptein (1). BCR/ABL causes chronic myeloid lekemia (CML) and some acute lymphoblastic lkemia cases [ALL (2)].
Imatinib reversibly inhibits the enzymatic ativity of Bcr/Abl, which is critical for its oncogenic potential (3, 4); therefore the block of B
Correspondence and reprint requests to: Carlo B. Gambacorti-Passe20133 Milan, Italy. Fax: 39.2.2390-3237. E-mail: gambacorti@istitut1 Department of Experimental Oncology, Istituto Nazionale Tumori,2 Section of Hematology, University of Milano Bicocca, S. Gerardo H3
361
Abl enzymatic (tyrosine kinase) activity represea rational strategy to treat BCR/ABL positileukemias, and a model for the developmeninnovative treatment modalities for cancer.
In preclinical models, imatinib manifestedpotent anti leukemic activity bothin vitro (5–7)and in vivo (5, 8). Inhibition of proliferationcolony formation and induction of apoptosis winduced by imatinib in BCR/ABL transformecells, as well as inhibition of Bcr/Abl autophophorylation. Initial clinical studies have co
epartment of Experimental Oncology, Istituto Nazionale Tumori, Viazian 1i.mi.it.nezian 1, 20133 Milan, Italy.l, Via Donizetti 106, 20052 Monza, Italy.
0
tions than the measurement of imatinib plasma le
Department of Oncology, Mario Negri Institute for Pharmacologica4 Department of Applied Bioscience, Swiss Federal Institute of Tech
002 Elsevier Science (USA)
arch, Via Eritrea 62, 20157 Milan, Italy., ETH Zentrum, CH-8092 Zurich, Switzerland.
firmed such a specific therapeutic activity (9, 10).These trials however, have also shown that BCR/ABL� patients fall into two categories regardingresponse to imatinib. Patients with CML inchronic phase experience durable responses, anda high percentage (�50%) of cytogenetic re-sponses; in this setting resistance to imatinib is arare phenomenon, at least at present. Patients withacute leukemias (blast crisis CML or Ph� ALL)initially experience high rates (�50%) of hema-tological responses, which however are soon fol-lowed by relapses in the majority (�80%) oftreated patients. In this latter group of patients,resistance to imatinib represents the dominantclinical and research issue at present. Patientswith accelerated phase CML fall in between thesetwo groups (11).
Resistance to imatinib was shown to derivefrom cellular mechanisms such as BCR/ABLgene amplification (12–14) or BCR/ABL muta-tions (15). In addition, in vitro and in vivo datahave shown that a plasma protein, �1 acid glyco-protein (AGP), binds imatinib with high affinity,alters imatinib bioavailability, and causes resis-tance to imatinib in an animal model (16).
In the present paper 12 patients in acceleratedphase or blast crisis CML, entered in 3 different clin-ical trials, were studied at the initiation of therapyand at relapse to assess the molecular mechanismsassociated to sensitivity or resistance to imatinib.
MATERIALS AND METHODS
Patients
Twelve patients affected by CML in blastcrisis or in accelerated phase were studied afterwritten informed consent. Nine patients were af-fected by myeloid blast crisis, and three by accel-erated phase CML. The patients were entered inthree different registrative or nonregistrative trialsperformed under Good Clinical Practice (GCP)guidelines. Imatinib was administered orally inone or two daily administrations. The total dailydosage of imatinib ranged between 400 and 800 mg.
Blood samples were obtained on day 1, im-mediately before treatment and subsequently dur-ing the first 14 days of treatment.
Heparinized (50 U/ml) samples were obtainedand treated as follows. Blood was centrifuged(400g for 30 min at 15°C) on a Ficoll gradient,and mononuclear cells recovered between the Fi-coll and the plasma layers. Cells were washedtwice in cold PBS (BioWhittaker Europe, Ver-viers, Belgium) and resuspended in RPMI � 10%fetal bovine serum (FBS). All samples analyzedcontained �60% blasts.
Chemicals
Imatinib was provided by Novartis PharmaAG, Basel, Switzerland. It is a derivative of a2-phenylaminopyrimidine, with a molecular weightof 590. For in vitro experiments, stock solutionsof imatinib were prepared at 1 and 10 mM indistilled water, filtered, and stored at �20°C.
Erythromycin was used as erythromycin base(Sigma Chemical Co., St. Louis, MO). erythro-mycin was dissolved in ethanol and then diluted1:1000 in distilled water and used.
Determination of the in Vitro Cell ProliferationActivity ([3H]Thymidine Uptake Assay)
Six to eight replicate cultures (200 �l), eachcontaining 104 cells, were incubated with 0–10�M imatinib in 96-well microtiter plates (CorningCostar Corp., Cambridge, MA) for 0–54 h at37°C. After this period, 20 �l of RPMI-1640medium (BioWhittaker Europe, Verviers, Bel-gium) containing 10% FBS and [3H]thymidine ata dose of 1 �Ci per well (DuPont NEN, Boston,MA) was added to each well. After an additional18 h, cells were harvested and transferred to afilter (Printed Filtermat; Wallac Oy, Turku, Fin-land). [3H]Thymidine uptake was measured in a1205 betaPlate liquid scintillation counter (WallacInc., Turku, Finland). The IC50 inhibitory concen-tration (defined as the concentration of STI571producing a 50% decrease in proliferation com-pared with untreated controls) was calculated.
Determination of Apoptosis by FACS Analysis
The annexin V/propidium apoptosis assay wasperformed as described previously (12). Briefly,cells were incubated with a mixture of fluores-
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
362
cein-conjugated annexin V and propidium iodide(Bender MedSystems, Austria) at room tempera-ture for 15 min in the dark, and immediatelyanalyzed by double fluorescence FACScan. Com-pensation was performed using samples labeledwith only annexin or propidium. Live cellsstained negative for both propidium and annexin(lower left quadrant).
Caspase 3-like Activity
Acetyl-Asp-Glu-Val-Asp-(amino-4-methylcou-marin) (DEVD-amc) hydrolytic activity was ana-lyzed as previously described (12). Cells werewashed in phosphate-buffered saline (PBS) andlysed in 100 �l of lysis buffer, containing 10mmol/L Hepes, pH 7.4, 0.1% Chaps, 2 mmol/LEDTA, and 2 mmol/L dithiothreitol (DTT). Pro-tein content was determined by the Bradfordmethod and volumes equivalent to 3 �g of proteinwere incubated in 500 �l reaction buffer contain-ing 20 mmol/L Hepes buffer, pH 7.4, 10% glyc-erol, 2 mmol/L DTT, and 20 �M DEVD-amc(Peptide Co., Kyoto, Japan) at 37°C for 2 h. Thereaction was stopped by the addition of 500 �lof reaction buffer without DEVD-amc. Sampleswere read immediately on a spectrophotofluorom-eter, at an excitation wavelength of 380 nm and anemission wavelength of 460 nm.
Western Blot Analysis
Cells were washed twice with ice-cold PBS andsubsequently lysed in 200 �l of 1� Laemmli’sbuffer (i.e., 50 nM Tris–HCl, pH 6.8, 2% SDS;0.1% bromophenol blue, 10% glycerol, 5% 2-mer-captoethanol), supplemented with phosphatase in-hibitors sodium vanadate (1 mM) and sodiumfluoride (10 mM). Cell lysates (from 2 � 105 to106 cells) were sonicated for 1 min, heated at95°C for 10 min, centrifuged, and then stored at�20°C. Proteins in the cell lysate were resolvedby SDS electrophoresis on 7.5 or 5% polyacryl-amide gels and transferred to Protran nitrocellu-lose membranes (Schleicher & Schuell, Dassel,Germany). Endogenous Bcr/Abl, tyrosine-phos-phorylated Bcr/Abl, and actin were detected withthe corresponding mouse monoclonal antibodiesor rabbit antiserum and visualized by enhanced
chemiluminescence detection (Amersham Corp.,Little Chalfont, UK) by use of horseradish perox-idase-linked goat anti-mouse or anti-rabbit immu-noglobulin G as the secondary antibody (Amer-sham Corp.). Anti-abl monoclonal antibody (cloneAb-3) was purchased from Calbiochem Corp. (LaJolla, CA). Anti-phosphotyrosine monoclonalantibody (clone 4G10) was purchased from Up-state Biotechnology (Lake Placid, NY). Rabbitanti-actin polyclonal antibody was purchasedfrom Sigma Chemical Co.
Densitometric analysis of Western blots wasperformed with an Eagle Eye 11 photodensitom-eter (Stratagene Cloning Systems, La Jolla, CA).Intensities of tyrosine-phosphorylated Bcr/Ablbands were quantified against standard dilutionsof patients’ samples (Fig. 1). Differences in theamount of Bcr/Abl protein loaded were similarlynormalized against actin levels.
HPLC-Based Determination of Imatinibin Human Plasma
Plasma samples (0.5 ml) were mixed with anequal volume of acetonitrile and kept at roomtemperature for 20–30 min. The protein precipi-
FIG. 1. Serial dilutions of lysates obtained from apatient sample (Pt 009). The upper panel shows Westernblots with anti-Abl and anti-phosphotyrosine mAbs. Thelower panel shows the densitometric analysis of the imagespresented in the upper panel.
Gambacorti-Passerini et al. Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
363
tate was removed by centrifugation at 13,000rpm � 5 min and 100 �l of the supernatant wasinjected into the separation module Alliance 2690(Waters, Milford, MA). HPLC analysis was car-ried out using a EC 125/4 Nucleosil 100-5 C18column in line with a CC 8/4 Nucleosil 100-5 C18precolumn (Macherey-Nagel, Duren, Germany).The column was equilibrated at the flow rate of 1ml/min with 10% v/v acetonitrile in water con-taining 0.05% trifluoroacetic acid (solution A)and then the sample was eluted using a gradient tothe final condition of 90% (v/v) acetonitrile inwater containing 0.05% (v/v) trifluoroacetic acid(solution B) over a period of 20 min. The columnwas prepared for the next sample by holding thiscondition for 5 min and then returning to theinitial condition that must be maintained for 5min. After chromatographic separation, peakswere detected at 270 nm using Waters 2487 ab-sorbance detector. The acquisition system was aMillennium32 software for chromatography (Wa-ters, Milford, MA). To prepare the calibrationcurve imatinib was added to blank human plasma,yielding a final concentration of 2000 ng/ml. Thissolution was further diluted in blank humanplasma to achieve analyte concentrations of 1000,500, 250, and 100 ng/ml. Standards samples wereprocessed as described above. The limit of detec-tion in plasma was 100 ng/ml.
To obtain micromolar concentrations, �g/mlvalues must be multiplied by a factor of 1.69.
Statistical Analysis
Student’s t test was applied to the data usingthe Prism software analysis program (GraphPad,San Diego, CA). P values of less than 0.05 wereconsidered to be statistically significant and werederived from two-sided statistical tests. All dataare presented as means � 95% confidence inter-val (CI). CI values are displayed when they ex-ceed 10% of the respective mean.
RESULTS
Patients
Table 1 presents the demographic data regard-ing the patients studied in the present report. Allpatients showed an initial response to treatment(decrease in WBC counts), although 2 patients didnot meet the criteria for objective response, asdefined in Table 1.
Initial in Vivo Effects of Imatinib
Several parameters were assessed, accordingto previously published in vitro data (6).
Spontaneous proliferation of mononuclear cellsobtained from patients was tested using a [3H]thy-
TABLE 1
Characteristics of Patients Studied
UPNType ofdisease
Phenotypeof blasts
WBC(day 1)
Clinical response[duration (days)] Relapse
Phenotypeof relapse
001 CML-BC M 42.4 No response Y M002 CML-BC Biphen 43.3 Partial response (42) Y L003 CML-BC M 87.91 Partial response (85) Y M004 CML-BC M 40.17 Partial response (139) Y M005 CML-BC M 8.55 Partial response (224) Y M006 CML-BC M 66.8 Partial response (175) Y M007 CML-BC M 123.12 Partial response (128) Y M008 CML-BC L 10.16 Partial response (46) Y L009 CML-BC M 44.5 No response Y M0502 CML-AP NA 18.9 Partial response (105) Y L0504 CML-AP NA 89 Partial response (602–ongoing) N NA0505 CML-AP NA 2.4 Partial response (67) Y M
Note. BC (blast crisis), �30% blasts; AP (accelerated phase), blasts between 15 and 29%; PR (partial response), reduction in the numberof blasts to �30% (BC) or �15% (AP), lasting at least 4 weeks, but without reaching normal blood counts and/or a normal BM status.
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
364
midine uptake assay. Spontaneous proliferationranged from 6700 to 21,850 cpm. Significant in-hibition of proliferation (�100% decrease relativeto day 1) was detected in 10 of 12 patients studied(range 130–900%). Proliferation inhibition wasevident as early as 24 h after imatinib administra-tion. The results from a representative patient areshown in Fig. 2 (day 1 vs days 2–8, P � 0.001 atall time points in comparison with day 1). Thispatient was particularly informative because ofthe homogeneity of the cell populations tested; infact the percentage of blasts in the peripheralblood remained fairly stable over the time ofstudy (14 days) at �80%, although the patient
experienced a dramatic decrease in the number ofwhite blood cells (from 102,000 to 2600).
Induction of apoptosis was studied using theAnnexin V binding assay.
Early apoptotic cells were defined as thosestaining positive for annexin V and negative forpropidium iodide (lower right quadrant); lateapoptotic or necrotic cells stained positive forboth annexin and PI (upper right quadrant). In-duction of apoptosis was considered positivewhen the percentage of apoptotic cells (rightlower plus right upper quadrants) doubled in com-parison with pretreatment samples. The resultfrom a representative patient are shown in Fig. 3.Early apoptotic cells increased from 11% (day 1)to 34% (day 4), with early � late apoptotic cellsvarying from 14 to 49% during the same period oftime. Eleven patients were evaluated with thistest; 9 showed increased early apoptotic cells dur-ing treatment with imatinib (range of increase inapoptotic cells 110 to 450%). Induction of apop-tosis was evident as early as 48 hours after thestart of the treatment.
In 3 patients the induction of apoptosis wasfurther evaluated by the detection of caspase-3-like activity in cell lysates. Figure 4 presents the
FIG. 2. Spontaneous proliferation of mononuclear cellsobtained from Pt 001. Day 1 identifies the sample obtainedimmediately before imatinib administration. Error bars rep-resent 95% confidence intervals.
FIG. 3. Induction of apoptosis in Pt 004 during imatinib treatment. The annexin V binding assay was used. Early and lateapoptotic cells are identified in the lower right and upper right quadrants respectively. Live cells are gated in the left lowerquadrant. “Day 1” identifies the sample obtained immediately before imatinib administration.
Gambacorti-Passerini et al. Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
365
data from one such patient, showing an inductionof caspase-3-like activity over baseline values(5–15 arbitrary units) by 24 h of treatment, with agradual decrease in the following days.
Bcr/Abl Phosphorylation
In six patients in whom a band compatiblewith Bcr/Abl could be identified in cell lysates,the analysis of Bcr/Abl phosphorylation statuswas performed. In all but one patient (504) theBcr/Abl protein could be visualized with bothanti-PTyr mAb and with anti-Abl mAb. In Pt 504anti-PTyr antibodies visualized a band migratingat a position compatible with Bcr/Abl (confirmedby the comigration of Bcr/Ablp210 isolated from aPh� leukemia cell line), while no signal wasobtained with anti-Abl mAb. In this patient theband visualized with anti-Actin Ab was used toevaluate the amount of protein loaded. Figure 5presents the results obtained in two patients. Theinhibition obtained was calculated at 44% forpatient 001 and 63% for patient 504. The averagemaximal inhibition obtained in the six patientswas 55.4 � 14%. The inhibition of Bcr/Abl au-tophosphorylation occurred rapidly, with maxi-mal levels achieved between days 2 and 8.
Effects at Relapse
Eleven of 12 patients relapsed during treat-ment. Relapses developed between days 21 and224; samples were obtained on the last day oftreatment before imatinib discontinuation. Evalu-able cell preparations were obtained from sevenpatients (001, 002, 004, 005, 008, 009, 505).
Imatinib Plasma Concentrations
Plasma concentrations were measured atsteady state (days 4–5) and at relapse. No sub-stantial variations were noted in both peak orminimum concentrations, AUC and half-life, al-though relapsed patients showed a trend towardhigher levels and longer half lives (Gambacorti, inpreparation). Therefore clinical relapses were notassociated with decreased plasma imatinib con-centrations.
Proliferation
Spontaneous proliferation of leukemic cellswas no longer inhibited, compared with pretreat-ment values. Figure 2 presents results from pa-tient 001, in whom cells obtained at relapseshowed no inhibition and even a significantlyhigher proliferation rate (day 1 vs day 23), despite
FIG. 5. In vivo inhibition of Bcr/Abl autophosphoryla-tion in Pt 001 and 504 following imatinib administration.“Day 1” identifies the sample obtained immediately beforeimatinib administration.
FIG. 4. Induction of caspase 3-like activity in themononuclear cells obtained from Pt 502. Day 1 identifiesthe sample obtained immediately before imatinib adminis-tration.
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
366
the fact that the patient imatinib plasma levels atthe time of blood drawing were 12.3 �M and theyranged on that day between 6.5 and 14.1 �M, asimilar lack of proliferation inhibition (cpm in therelapsed sample �50% of pre-imatinib samples)was observed in six additional patients.
Bcr/Abl Phosphorylation
In five patients the Bcr/Abl phosphorylationstatus at relapse could be evaluated. The resultsare presented in Fig. 6. In all five patients Bcr/Abl
was highly phosphorylated at relapse; the level ofautophosphorylation was similar or higher thanthat observed in the samples obtained before treat-ment, in spite of the presence of imatinib in theblood of all patients at concentrations higher than3 �M (range 4.8–20.1 �M).
Leukemic cells isolated from patients werealso incubated in vitro for 2 h with 3 �M imatinib,to assess their intrinsic sensitivity to imatinib. Inall five patients that could be tested, a complete ornear complete (�85%) inhibition of phosphory-
FIG. 6. Bcr/Abl autophosphorylation in five patients before imatinib administration and at relapse. For comparison, theinhibition obtained in vitro with 3 �M STI571 is shown. “Day 1” identifies the sample obtained immediately before imatinibadministration.
Gambacorti-Passerini et al. Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
367
lation was evident. This result was surprising,since all patients experienced plasma concentra-tions well above the 3 �M level at the time ofsampling.
These data suggest the presence of a block ofimatinib activity in the blood of the patients. AGPwas previously identified as a binder of imatinib,which can be blocked by excess erythromycin(16). Therefore unseparated blood samples fromtwo imatinib resistant patients (008 and 009) wereincubated with 100 �M erythromycin at 37°C for1 h before Ficoll separation. The results are shownin Fig. 7. The samples obtained from the patientsshowed a highly phosphorylated Bcr/Abl. In bothcases, incubation of unseparated blood with eryth-romycin before Ficoll centrifugation produced asubstantial inhibition of Bcr/Abl phosphorylation(74% for Pt 008 and 65% for Pt 009), close to thelevel of inhibition (�85%) obtained by incubatingthe isolated leukemic cells of the two patients invitro with 3 �M imatinib. The plasma imatinib
levels in these two samples were 8.5 �M (Pt 008)and 6.1 �M (Pt 009). Similar results were ob-tained in two additional patients (not shown).Since it is known that erythromycin does not exertany direct effect on Bcr/Abl autophosphorylation(16), these data indicate that the major part ofimatinib present in these two patients’ blood wasunable to enter cells because of AGP binding, andthat Erythromycin freed imatinib from AGP, en-abling it to enter cells and to inhibit Bcr/Abl. Thisfact explains why no in vivo inhibition of Bcr/Ablcould be observed in relapsed patients, in spite ofapparently therapeutic concentrations of imatinibin their plasma, and of a preserved in vitro sensi-tivity to the imatinib concentration measured inplasma.
In Vitro Sensitivity to Imatinib
In three cases (Pt 001, 004, and 005), leuke-mic blasts could be obtained, frozen, and success-
FIG. 7. Effect of erythromycin on Bcr/Abl phosphorylation in two imatinib-resistant patients. Unseparated blood sampleswere incubated with erythromycin (100 �M) at 37°C for 1 h. Mononuclear cells were then separated, lysed, subjected toSDS–PAGE, transferred, and hybridized with anti-Abl or anti Ptyr mAbs. Controls include blood samples incubated with PBS(left lanes) and mononuclear cells incubated in vitro with 3 �M imatinib after Ficoll separation (middle lanes).
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
368
fully thawed at both treatment initiation and re-lapse. Therefore, these three patients were testedfor in vitro sensitivity to imatinib. Samples ob-tained at relapse showed a relative resistance toimatinib, with the IC50 shifting from 0.01 to 0.05�M (Pt 001), 0.01 to 0.3 �M (Pt 004), and 0.03 to0.3 �M (Pt 005) (Fig. 8). We conclude that re-lapse was characterized by the selection of leuke-mic cells more resistant to imatinib. However, thelevel of resistance should have been theoreticallyovercome by the concentrations of imatinib ap-parently reached in vivo. These data indicate thatthe majority of blasts obtained from three re-lapsed, imatinib resistant patients were still sen-sitive to inhibition by imatinib at concentrationsapparently achieved in vivo. Therefore, plasma
imatinib levels do not reflect cellular/tissue con-centrations.
DISCUSSION
The scenario of treatment options for CMLand other Ph� leukemias is entering a new phase,thanks to the development of imatinib [Glivec(17)]. For the first time a specific and relativelynon toxic drug is able to induce remissions inmost patients with BCR/ABL� leukemias, and toobtain cytogenetic responses in the majority ofCML patients in chronic phase. However the verybrief follow up period does not permit a definitiveassessment of this drug. It is therefore essential toobtain not only longer follow ups, but also tounderstand the reasons for the successes and thefailures of imatinib.
The present work was focused on the failureof imatinib to eradicate leukemia in blast crisis/accelerated phase CML patients. In the 12 pa-tients tested, imatinib initially caused the samephenomena (block of proliferation, induction ofapoptosis, inhibition of Bcr/Abl autophosphoryla-tion) that were described in vitro. However theextent of Bcr/Abl phosphorylation inhibition ob-tained in vivo was substantially lower than thatobtainable in vitro with similar imatinib concen-trations.
When the patients relapsed, the biological ef-fects of imatinib were no longer present. In par-ticular, the phosphorylation of Bcr/Abl was nolonger inhibited by imatinib, in spite of stable andhigh plasma concentrations. When isolated leuke-mic cells obtained from relapsed patients wereincubated for a short time in vitro with imatinibconcentrations similar to those present in vivo, adramatic inhibition of Bcr/Abl autophosphoryla-tion was observed, as well as a complete inhibi-tion of cell proliferation (not shown). To furtherstudy such a phenomenon, unseparated bloodsamples from relapsed patients were incubated invitro with erythromycin, a drug known to com-pete with imatinib for binding to AGP. Underthese experimental conditions imatinib could en-ter cells and exert its biological activity. Thesedata indicate therefore that most imatinib presentin plasma is not biologically active. In vitro data
FIG. 8. Comparison of in vitro sensitivity to imatinib inmononuclear cells obtained from patients before imatiniband at relapse. Cells were frozen and stored in liquid nitro-gen. “Day 1” and “ relapse” samples were thawed simulta-neously and assessed for in vitro sensitivity to imatinib.*P � 0.05; **P � 0.01 compared to “day 1” values. “Day1” identifies the sample obtained immediately before ima-tinib administration.
Gambacorti-Passerini et al. Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
369
show that AGP, when added at physiologicallevels to imatinib containing cultures, decreasesintracellular imatinib concentrations by a factor of10 or more (Gambacorti, submitted). In addition,the coadministration of imatinib and drugs likeclindamycin, which is known to bind AGP andtherefore able to compete with imatinib for AGPbinding, led to a dramatic decrease in imatinibplasma concentrations [(18) and Gambacorti, sub-mitted]. These data indicate that following thedisplacement of AGP, imatinib rapidly distributesto tissues, according to a steep plasma/tissuesgradient; they also confirm that most plasma ima-tinib is not in equilibrium with tissues. The abovedescribed conditions also suggest that concentra-tions of imatinib inside cells are not reflected byconcentrations in plasma, where the presence ofAGP subtracts a substantial portion of imatinibfrom the equilibrium. Therefore the assessment ofthe kinase activity of Bcr/Abl in leukemic cells(measured as autophosphorylation or substratephosphorylation) could represent a more relevantmeasurement of imatinib activity in tissue thanthe assessment of imatinib plasma levels.
These data indicate that leukemic cells areprobably exposed in tissues to imatinib levelssubstantially lower than those measured inplasma.
The exposure to marginally active concentra-tions of an active drug represents the optimalconditions for the selection of resistant cells, asextensively demonstrated in the field of infectiousdiseases and cancer chemotherapy.
We do not postulate raising AGP concentra-tions as a direct cause of resistance to imatinib inpatients. Basal AGP concentrations in mice arelow (�1/4 human levels), at difference with hu-mans in whom basal AGP levels (400–1000 mg/ml) are already sufficient to bind a substantial partof imatinib. The higher AGP levels observed inpatients with advanced leukemia (19, 20) willprobably amplify such a discrepancy betweenplasma and tissues concentrations, but are un-likely to cause resistance directly.
Rather we propose that the low tissue concen-trations of the drug represent an important factorthat increases the likelihood of selecting resistantcells.
Our data indicate that leukemic cells obtainedat relapse were indeed less sensitive than theircounterparts sampled before imatinib reflecting aprocess of cellular selection. However the factthat cells from all three patients that could betested were still sensitive to imatinib concentra-tions that apparently were reached in plasma, pro-vides further evidence that these levels were notreached in tissues. The in vitro IC50 values (0.05–0.3 �M) in the 3 resistant patients tested probablyreflect the tissue concentrations against which leu-kemic cells were selected during treatment.
Several cellular mechanisms for resistance toimatinib have been described by our and otherlaboratories. Gene amplification has been de-scribed in both resistant cell lines (12–14) andrelapsed patients (15), and constitutes the firstmechanism of resistance identified. Cell lineswith BCR/ABL gene amplification do not showan absolute type of resistance, rather they presenta quantitative one, with the IC50 shifting by nomore than one log. Therefore, a logical strategycould be to use higher dosages of imatinib, espe-cially as it has an excellent and quite benigntoxicological profile. Since the lowest effectivedose of imatinib in chronic-phase CML patients(which show IC50 of 0.1–0.3 �M) is 300 mg, thedosage needed for acute leukemia patients harbor-ing resistant cells caused by this type of alteration,could be in the range of 2–3 g/day, if tolerated. Asimilar approach could probably be followed incases where resistance is traced to increased ef-flux of imatinib from cells and the ensuing de-crease in intracellular concentrations, or wheretissue concentrations of imatinib are substantiallylower than plasma levels.
Three recent reports mention the presence ofmutations within the catalytic domain of Bcr/Ablin cells obtained from relapsed, resistant CMLpatients (15, 21, 22). In the first report a T315Imutation was detected in 6 resistant patients of 11studied. This residue is essential in mediating thedocking of imatinib inside Bcr/Abl: it establishesone H-bond with imatinib and because of its smallmass, allows the docking of a secondary aminogroup present inside imatinib. The substitutionwith isoleucine abolishes the H-bond and createsa “bump” inside the STI571 pocket, rendering the
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
370
binding of imatinib virtually impossible (15). Asecond report analyzed relapsed patients at a sin-gle institution (21). A second mutation (E255V)was identified. The modeling of such a mutationshows that this residue does not contact imatinib,is solvent exposed, and the N-terminal moiety ofAbl comprising the first �-strand and the subse-quent loop covering the hydrophobic binding sitein which the 2-amino-4-pyrydinilpiridine moietyof imatinib is located becomes more unstable(data not shown). A third report identified onemutation (E255K) in 1 of 12 patients tested (22),with structural consequences similar to those ofE255V. No substantial benefit from increasedimatinib concentrations would be expected inthese cases, particularly with the T315I mutation.In this case therefore, the strategy to overcomeresistance could be directed at developing newinhibitors able to bind and inhibit the mutatedform(s). It is interesting to note that another Bcr/Abl inhibitor (PD180970) with an increase of oneorder of magnitude in binding affinity to Ablcompared to imatinib (2.2 nM vs 38 nM) wasrecently published (23), and could be tested incells harboring BCR/ABL mutations. However,modeling shows the pocket accommodating the2-aminophenyl moiety of imatinib is also occu-pied by the phenyl substituent of PD180970, sug-gesting that PD180970 should not bind to theT315I mutant (not shown).
More recent reports have further increased thenumber of mutations present in resistant patients(24, 25).
A recent report (26) questions the ability ofAGP, isolated from donors or CML patients, tobind imatinib. While these data are intriguing, anumber of observations render their applicabilityto the in vivo situation of imatinib treated patientsquestionable (27).
The circumvention of clinical resistance toimatinib will require the evaluation of all factorsinvolved in this phenomenon (gene amplification,mutations, marginally active tissue concentra-tions, increased cellular efflux . . .), as well as thedevelopment of a multi faced approach (higherimatinib dosages, combination with cytotoxicdrugs, development of new Bcr/Abl inhibitors) tothe treatment of BCR/ABL expressing leukemias.
ACKNOWLEDGMENTS
This work was supported in part by the Italian As-sociation for Cancer Research (AIRC), the Italian Min-istry of Health (Ricerca Finalizzata), the Jose CarrerasInternational Foundation, the Antonio Castelnuovo Founda-tion, and MURST COFIN 2001. We thank Dr. LoredanaSpadola for modeling studies.
REFERENCES
1. Sausville, E. A. (1999) A Bcr/Abl kinase antagonistfor chronic myelogenous leukemia: A promising pathfor progress emerges [editorial; comment]. J. Natl.Cancer Inst. 91, 102–103.
2. Sawyers, C. L. (1999) Chronic myeloid leukemia.N. Engl. J. Med. 340, 1330–1340.
3. Cortez, D., Kadlec, L., and Pendergast, A. M. (1995)Structural and signaling requirements for BCR-ABL-mediated transformation and inhibition of apoptosis.Mol. Cell. Biol. 15, 5531–5541.
4. Goga, A., McLaughlin, J., Afar, D. E., Saffran, D. C.,and Witte, O. N. (1995) Alternative signals to RAS forhematopoietic transformation by the BCR-ABL onco-gene. Cell 82, 981–988.
5. Druker, B. J., Tamura, S., Buchdunger, E., et al.(1996) Effects of a selective inhibitor of the Abltyrosine kinase on the growth of Bcr-Abl positivecells. Nat. Med. 2, 561–566.
6. Gambacorti-Passerini, C., le Coutre, P., Mologni, L.,et al. (1997) Inhibition of the ABL kinase activityblocks the proliferation of BCR/ABL� leukemic cellsand induces apoptosis. Blood Cells Mol. Dis. 23, 380–394.
7. Deininger, M. W., Goldman, J. M., Lydon, N., andMelo, J. V. (1997) The tyrosine kinase inhibitorCGP57148B selectively inhibits the growth of BCR-ABL-positive cells. Blood 90, 3691–3698.
8. le Coutre, P., Mologni, L., Cleris, L., et al. (1999) Invivo eradication of human BCR/ABL-positive leuke-mia cells with an ABL kinase inhibitor. J. Natl. Can-cer Inst. 91, 163–168.
9. Druker, B. J., Sawyers, C. L., Kantarjian, H., et al.(2001) Activity of a specific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis of chronicmyeloid leukemia and acute lymphoblastic leukemiawith the Philadelphia chromosome. N. Engl. J. Med.344, 1038–1042.
10. Druker, B. J., Talpaz, M., Resta, D. J., et al. (2001)Efficacy and safety of a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia.N. Engl. J. Med. 344, 1031–1037.
11. Talpaz, M., Silver, R. T., Druker, B., et al. (2001)Glivec (imatinib mesylate) induces durable hemato-
Gambacorti-Passerini et al. Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372
doi:10.1006/bcmd.2002.0526
371
logic and cytogenetic responses in patients with accel-erated phase chronic myeloid leukemia: Result of aphase II study. Blood, in press.
12. le Coutre, P., Tassi, E., Varella-Garcia, M., et al.(2000) Induction of resistance to the Abelson inhibitorSTI571 in human leukemic cells through gene ampli-fication. Blood 95, 1758–1766.
13. Weisberg, E., and Griffin, J. D. (2000) Mechanism ofresistance to the ABL tyrosine kinase inhibitor STI571in BCR/ABL-transformed hematopoietic cell lines.Blood 95, 3498–3505.
14. Mahon, F. X., Deininger, M. W., Schultheis, B., et al.(2000) Selection and characterization of BCR-ABLpositive cell lines with differential sensitivity to thetyrosine kinase inhibitor STI571: Diverse mechanismsof resistance. Blood 96, 1070–1079.
15. Gorre, M. E., Mohammed, M., Ellwood, K., et al.(2001) Clinical resistance to STI-571 cancer therapycaused by BCR-ABL gene mutation or amplification.Science 293, 876–880.
16. Gambacorti-Passerini, C., Barni, R., le Coutre, P., etal. (2000) Role of alpha1 acid glycoprotein in the invivo resistance of human BCR-ABL(�) leukemiccells to the abl inhibitor STI571. J. Natl. Cancer Inst.92, 1641–1650.
17. Goldman, J. M., and Druker, B. J. (2001) Chronicmyeloid leukemia: Current treatment options. Blood98, 2039–2042.
18. Gambacorti-Passerini, C., Corneo, G., and D’ Incalci,M. (2001) Roots of clinical resistance to STI-571cancer therapy. Science 293, 2163.
19. Larghero, J., Mahon, F., Madalaine-Chambrin, I., etal. (2001) Elevated levels of the plasma protein alpha1 acid glycoprotein in chronic myelogenous leukemiain blast crisis mediate pharmacological resistance toGleevec (STI571, imatinib) in vitro and are associatedwith primary resistance in vivo. American Society of
Hematology, 43th Annual Meeting, Orlando, Florida.Blood 98, 616a n, 2582.
20. le Coutre, P., Kreuzer, K., Na, I., et al. (2002) Deter-mination of �1 acid glycoprotein in patients with Ph�chronic myeloid leukemia during the first 13 weeks oftherapy with STI571. Blood Cells Mol. Dis. 28, 75–85.
21. Hochhaus, A., Kreil, S., Corbin, A., et al. (2001)Roots of clinical resistance to STI-571 cancer therapy.Science 293, 2163.
22. Barthe, C., Cony-Makhoul, P., Melo, J. V., and Ma-hon, J. R. (2001) Roots of clinical resistance to STI-571 cancer therapy. Science 293, 2163.
23. Dorsey, J. F., Jove, R., Kraker, A. J., and Wu, J. (2000)The pyrido[2,3-d]pyrimidine derivative PD180970 in-hibits p210Bcr-Abl tyrosine kinase and induces apop-tosis of K562 leukemic cells. Cancer Res. 60, 3127–3131.
24. von Bubnoff, N., Schneller, F., Peschel, C., and Duys-ter, J. (2002) BCR-ABL gene mutations in relation toclinical resistance of Philadelphia-chromosome-posi-tive leukaemia to STI571: A prospective study. Lancet359(9305), 487–491.
25. Branford, S., Rudzki, Z., Walsh, S., et al. (2002) Highfrequency of point mutation clustered within the ATPbinding region of BCR/ABL in CML and Ph-positiveALL patients who develop Imatinib (STI57) resis-tance. Blood 99(9), 3472–3475.
26. Jorgensen, H. G., Elliott, M. A., Allan, E. K., Carr,C. E., Holyoake, T. L., and Smith, K. D. (2002)alpha(1)-Acid glycoprotein expressed in the plasma ofchronic myeloid leukemia patients does not mediatesignificant in vitro resistance to STI571. Blood 99,713–715.
27. Gambacorti-Passerini, C., le Coutre, P., Zucchetti, M.,and D’ Incalci, M. (2002) Binding of imatinib by alpha1 acid glycoprotein. Blood 100, in press.
Blood Cells, Molecules, and Diseases (2002) 28(3) May/June: 361–372 Gambacorti-Passerini et al.doi:10.1006/bcmd.2002.0526
