International Journal of Cell Cloning 7: (1989)

Transcription

1 ~ ~ Original Paper International Journal of Cell Cloning 7: (1989) Limiting Dilution Anaiysis for Detection of Residual Leukemic Cells After Bone Marrow Combined Decontamination with Mafosfamide Followed by Merocyanine-540-Mediated Photosensitization Adorfo Porcellini", Maria Teresa Marchetti Rossi', Annunziata Manna 4 Giovanni Sparaventi ', Carlo Pmaglia 'Servizio di Ematologia, Medicina 11, Centro Trapianto Midollo Osseo, Cremona, Italy; bivisione di Ematologia, Ospedale Civile, Pesaro, Italy Key Words. Autologous bone marrow purging * Mafosfamide * Merocyanine-540 LDA Abstract. Human acute myelogenous or lymphoblastic leukemia cells of the K-562 and CCRF-SB lines were mixed with an excess of normal human bone marrow cells to simulate a leukemia remission marrow. The cell mixtures were then incubated in vitro with mafosfamide (AZ) followed by the photoreactive dye merocyanine-540 (MC). Treated cells (1 x 104) were seeded in microwell plates, and increasing numbers of the line used to contaminate the normal marrow were added. Treatment with AZ alone produced total elimination (i.e., 6 logs) of CCRF-SB cells, while addition of merocyanine-540 increased the cloning efficiency from 22% to 24.4%. After treatment of the K-562-contaminated cell mixtures with AZ, nearly 1.6 logs of K-562 acute myelogenous blasts were still present, whereas AZ purging followed by MC-mediated photosensitization resulted in 100% elimination of clonogenic cells. Moreover, the combined treatment caused an increase of the cloning efficiency from 37.3% to 6276, clearly indicating that cleansing by the two agents combined was more effective than treatment with one agent alone. Introduction The high relapse rate observed in autologous bone marrow transplant (ABMT) as compared to allogeneic transplant has been attributed to both lack of graftversus-leukemia disease and occult leukemic cells contaminating the bone marrow during remission. Correspondence: Adolfo Porcellini, M.D., Centro Trapianto Midollo Osseo, Servizio di Ematologia, Medicina II, Cremona 26100, Italy. Received December 4, 1988; provisionally accepted February 15, 1989; accepted for publication March 13, Press

2 Porcellini/Rossi/Manna/Sparaventi/Pazzaglia 224 Although evidence exists in animal models that 4-hydroperoxycyclophosphamide (4-HC) or mafosfamide (AZ) is successful in purging marrow [l, 21 while sparing the pluripotential stem cells [3-71, no definite proof has been obtained that in vitro treatment of the autologous harvest is associated with better survival. However, there are some indications suggesting that ex-vivo manipulation of autologous harvest might be linked with improved survival in patients autografted in second complete remission (CR) [ In light of this recent data, it would seem that the minimal residual disease in the autologous harvest may be responsible, at least in part, for the higher incidence of relapse seen after ABMT. Therefore, ex-vivo manipulation of the autologous marrow to reduce clonogenic cells may play a role in reducing the relapse rate following ABMT. In previous studies we explored the efficacy of AZ, merocyanine-540 (MC) and hypertherrnia (43"C), either alone [ll] or in combination [El, in removing the residual clonogenic cells from bone marrow cell suspensions contaminated with different concentrations of human leukemia cells (CCRF-SB, K-562). We have also shown that combined AZ and MC are not harmful (at least to a certain extent) to pluripotential stem cells [l2]. Our data showed when the original contamination of clonogenic cells, prior to purging, was 5%. AZ or MC alone was quite effective in eliminating CCRF-SB acute leukemia cells and K-562 acute myelogenous cells [Ill. With higher blast contamination levels (10 or l5%), Combined treatments of AZ with MC or hyperthermia were not able to decrease the clonogenic cell frequencies to acceptable levels [12]. In view of these aforementioned results, we investigated the efficacy of AZ and MC combined. Our investigation included cleansing clonogenic cells from cell mixtures in which the leukemic contamination was not far from the values conventionally defined in clinical practice as complete remission (i.e., 5%). To this end we used a highly sensitive in vitro clonogenic assay which we recently developed [ll, 13, 141; this allowed detection of as much as a six-log reduction of leukemic cells, with a sensitivity threshold of one clonogenic cell per 1 X 104 bone marrow cells. Materials and Methods Details of the preparation of human bone marrow mononuclear cell suspensions (HBMMC), established leukemia cell line suspensions (CCRF-SB and K-562) and purging agents (AZ and MC), have been described previously [ll]. Cleansing Procedures Figure 1 shows the experimental design of the purging procedures. To simulate a remission bone marrow from leukemia patients, 1 x lo6 leukemia cells of the CCRF-SB or K-562 lines were mixed with a 19-fold excess of HBMMC; the suspensions were adjusted to 2 x lo6 cellslml with T(3 199 medium (Seromed, FRG), placed in 12 ml plastic tubes and incubated for 30 min at 37 C in a water bath with 100 pg/ml of AZ (Asta Werke, Bielefeld,

3 Combined Bone Marrow Purging with AZ and MC 225 FRG). After incubation, the cell suspensions were placed in ice for 5 min to stop the reaction and then washed twice in RPMI 1640 (Seromed). The suspensions purged with AZ alone were then diluted for evaluation to 1 X lo5 cells/ml by limiting dilution analysis (LDA). Photosensitization with Merocyanine-540 (MC) The fractions of the AZ-treated suspensions set aside for further processing with MC (Eastman Kodak Co., Rochester, NY) were diluted to 4 X lo6 cells/ml with a-mem containing 15% fetal calf serum (FCS; Seromed). The suspensions were then placed in Petri dishes, and 15 pglml of MC dissolved in 50% ethanol was added. Clear polystyrene tissue culture dishes (60 mm X 15 mm) containing the cell suspensions were exposed to light for 90 min in a device previously described in detail [ll]. After exposure, the cell suspensions were replaced in 12 d plastic tubes and washed twice in a-mem containing 5% FCS. Limiting Dilution Analysis At the end of the incubation period (with AZ alone or with AZ followed by MCmediated photosensitization), 1 X lo4 cells from those treated were suspended in 100 pl RPMI 1640 and plated in microculture plates (NUNC, Roskilde, Denmark). Cells from the same line (CCRF-SB or K-562) used to contaminate the HBMMC were added in increasing numbers to the 1 X lo4 purged cells contained in the microwells and incubated for 13 days. The cultures were refed with fresh medium every third day. Twelve hours before harvesting, the cells were pulse-labeled with 3H-thymidine. The cells were then harvested by means of a Skatron apparatus (Skatron, Lierbyen, Norway), and incorporated thymidine was quantitated by scintillation counting. Statistical Analysis Leukemic cell numbers persisting in the suspensions after purging were determined by the Poisson distribution relationship between the leukemic cells added to each well and the percentage of negative wells. The maximum likelihood (ML) method was used to plot the regression lines of best fit; ML regression analysis, x-intercept values and confidence intervals (CI) were calculated according to the method of Cox and Hinkley [U]. Cloning efficiencies (CE), calculated as the slope of the regression line when the ordinate is the natural log of the percent nonresponders, were obtained using the Minuit computer program (C.E.R.N. Library, Geneva, Switzerland) in a VAX-750 computer. Results To evaluate the frequency of leukemic cells of the CCRF-SB and K-562 lines surviving in a co-culture of HBMMC after treatment with either AZ alone or AZ followed by MC, an LDA was adopted [ll]. Briefly, the mean 3H-thymidine uptake in the 24 wells in which no leukemia cells were added was 6,920 cpm. From this value a threshold 3 SD above the mean was established. Wells with 3H-thymidine incorporation greater than this threshold were scored as responders (positive), while wells with 3H-thymidine uptake less than this cut-off were scored as nonresponders (negative).

4 Porcellini/Rossi/Manna/Sparaventi/Pazzaglia 226 igrios H B M M C u. los CCRF - SB!:.:;:;:;: or... K :::I::?::.:.x.:: AZ 100pg/ml MEROCYANINE I 37OC 30 min - DlLUllON TO 4,106 /r/ I-,-iCCRF-SB or K-562 cells / *'* I seeded by LDA ~ ' cells "purged, O O D O O with A2 + MC 540 'HTdr 12 h prior to harvesting Harvesting after 13 days incubation Fig. 1. Experimental design of bone marrow purging procedures. Cell mixtures of 1.9 x lo7 HBMMC and 1 X lo6 K-562 or CCRF-SB leukemia cells are incubated first with AZ, and subsequently analyzed with the LDA method. Fractions of the AZ-treated cell suspensions are set aside for dilution to 4 X lo6 cells/ml and further processing with MC (for details see Materials and Methods section). 1 X 1O4,mixture cells purged with either AZ alone or AZ + MC are then seeded in microwell plates to which increasing numbers of the same leukemic cells used to contaminate the HBMMC suspensions are added. With the addition of successively larger numbers of leukemic cells to the wells containing purged suspension, the proportion of responders increased in linear fashion (Fig. 2). This figure depicts the data obtained from LDA of HBMMC and CCRF-SB or K-562 leukemic cell mixtures either after purging with AZ alone, or with AZ followed by MC-mediated photolysis. The number of leukemic cells still present in the cell mixtures after cleansing is represented by the-x-intercept of the ML regression line, while the CE of CCRF-SB and K-562 cells was derived from the slope of the regression line. Figure 2 (upper panel) shows the results obtained by purging the cell mixtures contaminated with K-562 leukemic cells. With this leukemia cell line, the calculated residual clonogenic cells in each well were (95% CI to 0) after treatment with AZ alone, and 0 (95% CI-0.39 to 0) after purging with AZ and MC. The CEs of leukemic cells in this experiment, calculated from the slope of the ML regression line, were 37.3% and 62%, respectively. Two replicate experiments provided similar results: indeed, all K-562 blasts

5 Combined Bone Marrow Purging with AZ and MC 227 Fig. 2. Semi-log plot of the proportion of nonresponding microcultures for each dose of leukemic cells added to wells containing 1 X lo4 cleansed cell mixtures contaminated with 5% K-562 (upper panel) or CCRF-SB (lower panel). Results were obtained from LDA of cell mixtures decontaminated with AZ or AZ + MC-mediated photosensitization. The best fitting lines (as determined by the maximum likelihood regression analysis), x-intercept values, CE of clonogenic cells and confidence limits on the population regression line, were originated by the Minuit computer program using a VAX-750 computer. were completely cleared, not only by the combined treatment, but even by AZ alone. However, the CEs in the two experiments purged with only one agent were 21% and 18.6%, respectively, while they increased to 37.3% and 28.2%, respectively, in the two experiments in which the cell mixtures were purged with both agents. Again, the 95% CI confirmed that there was a 5% probability of the residual blasts being present at a frequency greater than one per 1 x lo4 purged mixture cells. In the experiment shown in Figure 2 (lower panel), complete decontamination of CCRF-SB cells in both cell mixtures, incubated with either AZ alone or AZ followed by MC, was obtained. As the figure shows, the CE of CCRF-SB cells in this purging experiment with AZ alone was 22%, increasing to 24.4% in the purging experiments with both agents. The 95% CIS for the value of the xlintercept of these experiments were to 0 after cleansing with AZ alone and -0.5 to

6 Porcellini/Rossi/Manna/Sparaventi/~z~glia after cleansing with both agents. In other words, there was less than 5% probability that the frequency of leukemic cells would be greater than 1 per 1 x 104 HBMMC in an assay in which the threshold sensitivity was about 1 clonogenic cell per 1 x ~O~HBMMC. These results were confirmed in two replicate experiments in which complete clearing of all leukemic cells was obtained even after treatment with AZ alone. In all these replicate studies, the 95% CI indicated that there was always a 5% probability that the frequency of residual blasts, after purging with either treatment, would be >1 per 1 x lo4. The CE in the replicate experiments was 7.7% and 18.8% (AZ), and 20% and 17% (AZ plus MC-540), respectively. Discussion Although we still lack definite proof that ex-vivo purging of the autologous marrow harvest is linked with better survival, there is a body of evidence suggesting this hypothesis to be true [8-lo]. The ideal agent for ex-vivo bone marrow purging should meet the requirements of both selectively killing residual leukemic cells contaminating the autologous harvest and sparing a sufficient number of hemopoietic stem cells to ensure hemopoietic reconstitution. Previous studies have shown that in experimental models, marrow cell suspensions treated with 4-HC, AZ or MC retain their ability to restore hemopoiesis [3-6, 161. Also, evidence has already been presented which suggests that the drug combination AZ plus MC used in these experiments is not harmful to pluripotent stem cells [El. Quantitation of the number of clonogenic cells present in the remission marrow along with estimation of those removed by ex-vivo purging still remains a major obstacle. Although differing somewhat from the clinical situation, in vitro simulation experiments may be useful for evaluating the efficacy of the drug or drug combination tested for killing leukemic cells. To this end we have adopted an LDA as described previously [11]. In all the experiments performed in which the ML regression lines are consistent with single-hit kinetics (Fig. 2). it is assumed that the test-mixture cell suspensions are homogeneous; therefore, the number of leukemic cells will occur in each sample according to the Poisson distribution equation. The 0 term of this equation will then indicate that the proportion of negative wells is a negative logarithmic function of the number of leukemic cells added to each well. Hence, the leukemic cell frequency in each sample can be calculated from the ML regression line which best fits. In Figure 2, the leukemic cells added to each well are plotted against the number of negative wells. Therefore, the regression line will be expected to include the origin only in the case in which the cleansing procedure was 100% effective (i.e., for 0 leukemic cells added, 100% negative wells are obtained). In any other case in which the cleansing was not complete, the leukemic cells still present in the wells after purging give rise to some positive wells even when no leukemic cells are added to the wells (i.e., for 0 leukemic cells added, a fraction

7 Combined Bone Marrow Purging with AZ and MC 229 of positive wells is obtained); thus, semi-log plots of leukemic cells added and the proportion of non-responders will not include the origin. In this case, the x-intercept will represent the number of leukemic cells still present in the wells after purging or, in other words, the number of leukemic cells that would have to be removed from the cell mixtures in order to achieve 100% nonresponders. In our in vitro clonogenic model, after purging with AZ, as few as acute leukemia K-562 cells were detectable in the cell mixtures; combined treatment with both AZ and MC not only depleted all leukemic cells, but also increased the CE, by a factor of 1.67, to 62%. In the CCRF-SB-contaminated mixtures, 100% cleansing was obtained with each purging protocol. Thus, the efficacy of the two treatments cannot be compared on the basis of the number of remaining leukemic cells. However, in the experiment shown in Figure 2, an increase from 22% to 24.4% was observed, indicating that cleansing by the two agents combined was more effective than treatment with a single agent. In fact, in our microtray assay system, an excessive cell proliferation rate can occur with a greater than expected proportion of negative wells, due either to a high number of residual leukemic cells remaining after purging or because too many cells were seeded per well. These false-negative results are due to the fact that if excessive cell proliferation occurs, medium supply may become a limiting factor in spite of the fact that the microcultures are refed periodically. In addition, as we have already pointed out, the threshold limit for detection with our LDA is 1 leukemic blast per 1 X lo4 HBMMC, i.e., 0.01%. Thus, a number of clonogenic cells that escaped the purging treatment, but were not detected by our LDA, could account for some false-negative results, and, in turn, a gentler CE slope than might be expected. It is possible that more aggressive purging (i.e., combined treatment rather than a single agent) may result in improved cleansing of residual clonogenic cells, albeit beyond the detection level of our assay. The fewer false-negative wells that would result would produce a higher CE. Therefore, the CE can be an important parameter for judging the efficacy of the cleansing procedure. This drug combination, which in an experimental model has previously been shown to spare enough pluripotent stem cells to ensure hemopoietic reconstitution [U], has also proved highly efficient in killing established leukemia cell lines as compared to treatments with the same agents used alone. The kinetics of the exponentially growing leukemic cells, such as those used in this model, may not be comparable to that of occult malignant cells contaminating the remission bone marrow, which may prove more resistent to the in vitro active agents. It is tempting to suggest that the combination studied in these experiments may be more effective, not only in experimental models which utilize leukemia cell lines, but also in clinical practice (especially as MC is active on leukemic cells regardless of their proliferation rate, and leukemic cells have not previously developed any pharmacological resistance to impermeant photoreactive dyes, such as MC).

8 Porcell ini/rossi/manna/sparaventi/pazzaglia 230 Acknowledgments The authors are deeply indebted todr. Giunrii Antonioli of the Department of Physics, University of Parrna, for his invaluable assistance with statistical analysis and to Angus Duwson for his help in preparing the manuscript. Supported in part by the Progetto Finalizzato Oncologia of the Italian National Research Council (C.N.R.) with Grant No and This manuscript was presented in part at the Fourth International Symposium on Autologous Bone Marrow Transplantation, August 19, 1988, Houston, Texas, and published in abstract form in the proceedings of the meeting. References I Sharkis SJ, Santos GW, Colvin M. Elimination of acute myelogenous leukemic cells from marrow and tumor suspensions in the rat with 4-hydroperoxycyclophosphamide. Blood 1980 $5: Sarpol SC, Epstein RB. Chemoseparation of cell populations in dogs by activated cyclophosphamide. Blood 1979;54:230a. Porcellini A, Manna A, Talevi N, et al. Effect of two cyclophosphamide derivatives on hemopoietic progenitor cells and pluripotential stem cells. Exp Hematol 1984; 12~ Rowley SC, Sharkis SJ, Hattemburg C, Sensenbrenner L. Culture from human bone marrow of blast progenitor cells with an extensive proliferative capacity. Blood 1987; 69: Degliantoni G, Mangoni L, Rizzoli V. Normal blast colony formation: an in vitro tool for monitoring human bone marrow purging. Bone Marrow Transplantation 1986;1: Gorin NC, Douay L, Laporte JP, et al. Autologous bone marrow transplantation using marrow incubated with Asta Z 7557 in adult acute leukemia. Blood 1986; Sahovic EA, Colvin M, Hilton J, Ogawa M. Role for aldehyde dehydrogenase in survival of progenitors for murine blast cell colonies after treatment with 4-Hydroperoxycyclophosphamide in vitro. Cancer Res 1988;48: Yeager AM, Kaiser H, Santos GW, et al. Autologous bone marrow transplantation in patients with acute nonlympholeukemia, using ex vivo marrow treatment with 4-hydroperoxycyclophosphamide. N Engl J Med 1986;315: Gorin NC, Aegerter P, Auvert B. Autologous bone marrow transplantation (ABMT) for acute leukaemia in remission: fifth European survey. Evidence in favour of marrow purging. Influence of pretransplant intervals. Bone Marrow Transplantation 1988;3(suppl 1): Mangoni L, Carella A, Aglietta M, et al. Autologous bone marrow transplantation (ABMT) for acute lymphoblastic leukemia in remission using marrow treated with mafosfamide. Blood 1987;70: (suppl 1):322a. Porcellini A, Talevi N. Marchetti-Rossi MT, et al. Limiting dilution analysis for the determination of leukemic cell frequencies after bone marrow decontamination with mafosfamide or merocyanine-540. Blood 1987;70: Porcellini A, Manna A, Marchetti-Rossi MT, et al. Experimental basis of bone marrow purging with chemical and photodynamic agents. In: Baum SJ, Santos GW, Takakt~ F, eds. Recent Advances and Future Directions in Bone Marrow Transplantation. Experimental Hematology Today. New York: Springer-Verlag, 1988:45-48.

9 Combined Bone Marrow Purging with AZ and MC Moretta A, Pantaleo G, Moretta L, Ceroaini JC, Mingazi MC. Direct demonstration of the clonogenic potential of every human peripheral blood T cell. J Exp Med 1983; 157: Martin PJ, Hansen JA. Quantitative assays for detection of residual Tells in depleted human marrow. Blood 1985;65: Cox DR, Hinkley DV. Theoretical Statistics. 1st edition. London: Chapman and Hall, 1974 : Sieber F, Meagher RC, Spivak JL. Differential sensitivity of mouse hematopoietic stem cells to merocyanine 540. Differentiation 1981;19:65-68.