in vitro growth modulation by l-ascorbic acid of colony ... · 51 patients with mds, received from...

[CANCER RESEARCH 52, 4458-4466, August 15, 1992]

In Vitro Growth Modulation by L-Ascorbic Acid of Colony-forming Cells fromBone Marrow of Patients with Myelodysplastic Syndromes1

Chan H. Park,2 Bruce F. Kimler, David Bodensteiner, Sean R. Lynch, and Ruth S. Hassanein

Departments of Medicine [C. H. P., D. B., S. R. L.], Radiation Oncology [B. F. K.J, and Biometry fR. S. H.], University of Kansas Medical Center. Kansas City,Kansas 66160

ABSTRACT

In vitro colony growth was studied on bone marrow cells from 51patients with myelodysplastic syndromes (MDS), using a cell culturemethod with the unique feature of daily feeding, in an effort to gaininsight into the pathophysiology of MDS and to assess the clinicalutility of this cell culture assay. The colony growth pattern of MDSmarrow cells is remarkably similar to that of acute myeloid leukemia butquite dissimilar from that of normal marrow, in support of a commonpathophysiological mechanism for these two disorders. In particular,i -ascorbic acid (LAA) enhanced colony growth in 30% and suppressedgrowth in 16% of cases, a finding also similar to that in acute myeloidleukemia, indicating a unique growth requirement which may be explored for therapeutic purposes. Further, these LAA effects have prognostic value, with LAA-sensitive (both LAA-enhanced and LAA-sup-pressed) cases displaying shorter survivals than LAA-insensitive cases(median survival of 5 months versus 18 months; P = 0.011). This prognostic value is independent of, and more powerful than, bone marrowblasts; the median survival was 18 months for <5% bone marrow blastsand 8 months for >5% bone marrow blasts (P = 0.044). These two riskfactors can be used together to identify patients with an extremely goodor an extremely poor prognosis. This study establishes the clinicalusefulness of the LAA effect in MDS as a prognostic factor and providesa new lead to explore in understanding differential biochemical/molecular events and, possibly, a new therapeutic approach to the management of MDS.

INTRODUCTION

LAA3 has been known for 20 years to promote the growth of

mouse plasmacytoma and human myeloma cells in vitro (1,2).Using a unique cell culture method employing daily feeding ofnew medium, we demonstrated that LAA also modulated the invitro growth of leukemic colonies from patients with AML(3-5). These growth-modulatory effects of LAA have recentlybeen reviewed (6). A similar result has been reported by Alcainet al. (7) for a human promyelocytic leukemia cell line, usingascorbate and its free radical. The daily feeding culture systemwas used here to study colony growth of bone marrow cells frompatients with MDS, preleukemia, in an effort to gain insightinto the pathophysiology of MDS and to assess the clinicalvalue of this culture assay. This report describes the result thatthe colony growth pattern of MDS is remarkably similar to thatof AML but dissimilar from that of normal controls, in supportof a common pathophysiology for these two disorders. Especially notable is that LAA can modulate the colony growth ofMDS much like AML. This LAA effect was shown to be bio-

Received 2/21/92; accepted 6/4/92.The costs of publication of this article were defrayed in part by the payment of

page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

'Supported by a grant (ROÃ• CA 20717), a Research Career DevelopmentAward to C. H. P. (K04 CA 00534) from the NIH, and the A. S. and Minnie Q.Underwood Fund.

2 To whom requests for reprints should be addressed, at Division of Oncology/Hematology, Department of Internal Medicine, Texas Tech University HealthSciences Center, Lubbock, TX 79430.

3 The abbreviations used are: LAA, L-ascorbic acid; AML, acute myeloid leukemia; FAB, French-American-British; MST, median survival time; MDS, myelodysplastic syndromes.

logical in nature when AML was studied (5). The effect in thisstudy was shown to have a clear clinical implication, with apowerful prognostic value.

MATERIALS AND METHODS

Patients and Bone Marrow Samples. Bone marrow aspirates from51 patients with MDS, received from the University of Kansas MedicalCenter and three referral hospitals in the Kansas City area, between1978 and 1987, were the subject of this study (Table 1). All patients hadprimary refractory anemia with normocellular or hypercellular bonemarrow. There were also varying degrees of dysmyelopoiesis. Patientswith marrow blast cells of >30% were excluded, in compliance with thecriterion set for MDS by the French-American-British CooperativeGroup (8). Normal marrow samples used as controls were obtainedfrom hematologically normal individuals as part of staging for othermalignancies and from bone marrow transplantation donors. No patient had received cytotoxic chemotherapy prior to the time of bonemarrow aspiration for this study. Informed consent was obtained fromall patients, as designed and approved by the University of KansasMedical Center Human Subjects Committee. All patients were evaluated by a hematologist/oncologist, and all bone marrow samples wereexamined by a hematologist and/or pathologist at the time of bonemarrow procurement for this study. Later, two hematologists (D. B.and S. R. L.) reviewed (and confirmed the diagnosis of MDS for) allbone marrow slides. The follow-up information on survival was sought

on all patients from the consultant hematologist/oncologist and/or theprimary referring physicians. Survival duration was calculated from thedate of bone marrow aspiration by which the diagnosis was made.

Storage of Cells. Bone marrow buffy coat cells were frozen in asmany aliquots as possible. A solution of 10% each of dimethylsulfoxide(Sigma, St. Louis, MO) and fetal calf serum (Flow Laboratories, Rock-ville, MD) in LAA-free a medium (GIBCO, Grand Island, NY) wasused to suspend cells, which were frozen at a slow rate (TC/min). Vialswere then immersed and stored in liquid nitrogen (â€”180'C) until use. A

20-30% reduction of viability was observed with the freezing process,but long-term storage was not associated with any additional loss ofcells. For the sake of technical uniformity, all samples (MDS and normal) were frozen once before being assayed.

Cell Culture Assay. A detailed description of the cell culture methodused in this study was reported previously (9). Briefly, the cell culturesystem consisted of two layers of 0.3% agar in a 35-mm plastic Petridish that was perforated on the bottom with five small holes and storedwithin a 60-mm Petri dish for sterility. Stored cells were thawed rapidly, washed, and incorporated into the top agar layer. Both layerscontained a growth medium consisting of 70% LAA-free a medium,15% fetal calf serum, and 15% leukocyte-conditioned medium. Thelatter was prepared by incubation of normal human peripheral leukocytes with phytohemagglutinin (Wellcome Research Laboratories,Beckenham, England).

From identically inoculated plates for each patient or normal control, three groups of dishes were randomly assigned. Starting on the dayof inoculation, two groups of five dishes each received daily feeding ofthe growth medium (0.5 ml), with one of these also receiving LAA(Sigma) added to the feeding medium at 0.3 min (a normal physiologicalconcentration). A third group of dishes did not receive daily feeding.Since LAA is known to have an extremely short half-life in culture (10,11), LAA was added every day, with daily feeding, to the group receiving LAA. Glutathione was also added, at 0.3 min, whenever LAA was

4458

on March 28, 2021. © 1992 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GROWTH MODULATION BY L-ASCORBICACID

Table 1 Patient data: clinical and clonogenic assay variables

Pt.123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051ivrAge(yr) MDS" BM(%)*85

I'2612'268

T276T8556528176765756232666867758190667461733575667657788075792361418755187679686768814873765079657723121511030281922012612430243162101111171224161522111411698252272271012154FAÃŸrank11455331355131333315343355321353333553353331Chromosome

aberration'NormAbnAbnAbnNormNormAbnAbnAbnAbnAbnAbnAbnAbnNormNormAbnAbnAbnAbnAbnNormChemotherapyNNNYNNNNNNYNYNNNNYYNYNNNYYNYNNNNNNYNNNNNYNYNNNYNYYNLeukemiaprogressionNNNYYNNNYNYYYNYYNYYYYNNNYNYYYYNNNNYYNNNNYYYNYNYNFeedingratio<3<3Absolute>3>3>3Absolute<3>3>3Absolute<3<3<3Absoluteâ€”>3<3<3>3AbsoluteAbsoluteâ€”â€”>3â€”<3â€”<3>3->3<3<3â€”Absolute-<3<3>3â€”â€”â€”>3â€”<3>3<3<3<3-LAAratio1.070.840.420.990.892.040.260.910.690.992.70.710.640.801041.042.081.133.40^1.332.050.700.637.671.460.930.660.640.921.212.157.420.841.080.591.00Ttest

Pvalue0.100.200.100.200.200.050.050.200.100.200.050.010.100.100.0010.200.050.200.010.0010.200.200.200.050.0010.200.200.050.200.200.100.010.0010.200.200.051.00Survivaltime

(months)1*14+66+12018184423+13620+43+102041123217332230+63210+821454+19151341

+20132192223738133719

" 1", primary; 2Â°,secondary.* BM, bone marrow blasts.c Norm, normal; Abn, abnormal.d+, patient alive.

added. Glutathione was shown to potentiate the effect of LAA but wasineffective by itself (1, 3, 4). The fed medium diffused through the agarlayers and drained out of the dish through the holes in the bottom.

Cultures were incubated for 2 weeks or longer at 37 Â°Cin an atmo

sphere continuously flushed with 7% CO2. Colonies of 50 or more cellswere counted, usually at 2 weeks of culture, using an inverted microscope. In the minority of patients in whom colony growth was slower,colonies were counted a few days later but always before 3 weeks ofculture time. However, for any given patient, all counting was completed within a 24-h period. It has been previously substantiated thatcolonies grown in this culture system from bone marrow of leukemicpatients are leukemic in origin (3, 4, 9, 12), while normal bone marrowproduces normal myeloid colonies (granulocyte/macrophage colony-forming cells) (13). The initial cell number plated per dish was 5 x IO5,

but, in repeat studies, this was often appropriately reduced in patientswith high plating efficiencies. Throughout this study period, there wasessentially no change in cell culture methodology or other laboratorytechniques.

Interpretation of Results. A simple / test was used to determinewhether, within an individual experiment, there was a significantdifference (P < 0.05) in the number of colonies formed with the addition of LAA versus when LAA was not added. Alternatively, specimenswere classified as either LAA-sensitive or LAA-insensitive, dependingon the value of the LAA ratio, obtained when the number of coloniesproduced in the presence of LAA was divided by the number formedwhen LAA was not added.

Statistical Analysis. Standard t tests and x2 tests were used for com

parison of sets of quantitative data (e.g., number of colonies producedwith and without LAA or percentage of bone marrow blasts in LAA-sensitive and LAA-insensitive cases) and qualitative data (number oflive patients among LAA-sensitive and LAA-insensitive populations).Comparisons of survival durations were examined by the univariateGehan-Wilcoxon test for singly censored samples (Number ChruncherStatistical System, Version 5.5, Dr. Jerry L. Hintze, Kaysville, UT). ACox proportional hazard model (BMDP 2L, BMDP Statistical Software, Inc., Los Angeles, CA) was utilized to identify independent

4459



GROWTH MODULATION BY L-ASCORBIC ACID

predictors of survival. Finally, both stepwise logistic regression (BMDPLR) and stepwise discriminant analysis (BMDP 7M, BMDP StatisticalSoftware) were used to identify significant independent predictors forclinical progression to leukemia.

RESULTS

The total number of MDS cases for which bone marrowsamples were available for cell culture studies was 51 (Table 2).Of these, 38 (75%) grew 10 or more colonies per 5 x 10s cells

inoculated. Among these 38, 17 cases had colony numbers between 10 and 30 and 10 cases had colony numbers greater than1000. Therefore, 27 cases (71%) had colony numbers clearlyoutside the normal range, i.e., 50-597 colonies shown with 24normal controls during the same time period. The standarddaily feeding increased the colony number for MDS by >3-foldin 20 (53%) cases, with seven (18%) cases demonstrating anabsolute necessity of daily feeding for colony growth. In contrast, colony growth of normal bone marrow was not significantly affected by feeding, with no cases showing colony number increased >3-fold by feeding.

Of the 37 MDS specimens from which colonies were formedand in which the LAA effect was tested (Table 2), a number ofcases demonstrated an increase in colony number, while forothers there was a reduction in colony number, with the addition of LAA (Fig. 1). Among 24 normal bone marrow specimens, none had a significantly higher number of colonies withLAA, while only three (13%) had marginally lower numbers(Fig. 1). Repeat cultures of the same bone marrow aspirates,performed multiple times and with simultaneous normal controls, confirmed that these LAA effects, either enhancement orsuppression, were selective for MDS and consistent with theinitial culture results.

Using a t test to detect a significant difference (P < 0.05) inthe number of colonies formed, 14 specimens (38%) exhibited asignificant effect with the addition of LAA versus when LAAwas not added. Specifically, colony numbers were increased fornine specimens (24%) and decreased for five (14%) by the addition of LAA (Table 2). It was observed that, for the most part,specimens that displayed a significant difference in colony number (by t test) exhibited an LAA ratio of either <0.65 or >1.35.Using this more quantitative delimiter of the effect of LAA, itwas found that 17 (46%) specimens were LAA sensitive, with

Table 2 Distributions according to clonogenic response

10.0

GroupTotal

populationNo

colonygrowthColonygrowthdemonstratedColony

number >10 and<30Colonynumber>1000Colonynumber outside normalrangeDaily

feeding increasedcolonynumber>3-foldDaily

feeding absoluterequirementLAA

effect tested/defined by PvalueLAA-insensitiveLAA-sensitiveLAA-suppressedLAA-enhancedLAA

effect tested/defined byratioLAA-insensitiveLAA-sensitiveLAA-suppressedLAA-enhancedNumber

(%)51

(=100)13(25)38

(75)17102720737231459372017611(=100)(45)26(71)(53)(18)(=100)(62)(38)(14)(24)(=100)(54)(46)(16)(30)(=100)(36)(64)(-100)(35)(65)

o

O-C'Ã¬

_c

ODi

OU

O)nE

1.0

0.2

Â¡1â€¢¿�

Normal MDSFig. 1. Effect of LAA on growth of normal and MDS colony-forming cells. The

ratios between the numbers of colonies with LAA and without LAA are shown.Arrows, values above the indicated limit. Normal and MDS experiments shownwere done during the same period (November 1980 through March 1987), although not necessarily performed simultaneously.

Table 3 Survival analysis by clonogenic responses

GroupCloning

assayperformedNocolonygrowthColony

growthdemonstratedDailyfeeding effect of<3-foldDailyfeeding effect>3-foldDailyfeedingrequiredLAA

effect"LAA-insensitiveLAA-sensitive

LAA-suppressedLAA-enhancedLAA

effect*LAA-insensitiveLAA-sensitive

LAA-suppressedLAA-enhancedNumber511338181372314

592017

611MST

(months)13171313711166

46185

47T_jnr<-â€¢l1!_T1up0.890.220.910.038

0.270.0590.011

0.190.022value1-0.363-0.933"0.88

" Defined by P value of <0.05 for individual t tests.* Defined by ratio of number of colonies with or without LAA.

six (16%) showing suppression (LAA ratio <0.65) and 11(30%) showing enhancement (LAA ratio >1.35).

There were only seven instances where there was not complete agreement between the two methods of defining LAAsensitivity. For the 20 patients in the range of LAA ratiodefined as LAA insensitivity, all but two showed no statisticallysignificant difference in colony numbers with and without LAA(P > 0.05, based on t tests within individual experiments). Thetwo patients that did show a difference yielded ratios of 0.66and 0.71, near the lower boundary of the range. For the 17patients that fell outside the LAA-insensitive range and were,

therefore, termed LAA sensitive (either suppression with a ratioof <0.65 or enhancement with a ratio of >1.35), five did notshow statistically significant differences in colony numbers withand without LAA (P > 0.05), primarily because some disheswere lost due to contamination and the power of the t test was,therefore, not high enough to confirm a statistical difference.To avoid a situation where simply the number of plates contaminated might affect the categorization of a particular patientas either LAA sensitive or LAA insensitive, the LAA ratioapproach was preferred, since it utilized a quantitative end-point. However, both methods of defining LAA effectivenesswere used in subsequent analyses.

4460




We next addressed the question of whether these clonogenicassay results had prognostic value. Patient survival was evaluated according to the results of the clonogenic assay (Table 3).Patient survival data were available as follows: 42 dead, threecensored for death due to other causes, three lost to follow-up,and three still alive. The MST for all 51 patients was 13 months(Table 3). This is in general agreement with published data(14, 15). Although those patients that did not exhibit colonygrowth survived somewhat longer (MST = 17 months) than didthose for whom colony growth was demonstrated (MST =13months), the difference was not statistically significant (P >0.25), as analyzed by the Gehan-Wilcoxon test. Likewise, thedaily feeding effect was not of prognostic value. Median survival was not predicted (P > 0.20) by whether there was a<3-fold increase in colony number, a >3-fold increase, or evenan absolute requirement for LAA.

Turning to the question of whether growth modulation byLAA was of predictive value, we first analyzed the survival dataaccording to whether there was LAA sensitivity, as defined by astatistically significantly different number of colonies formedwith and without LAA addition (LAA effect, by t test). Thosepatients whose cells were insensitive to modulation by LAAshowed a greater MST (16 months) than did those patients forwhom colony growth was either suppressed (MST = 4 months)or enhanced (MST = 6 months). However, because of smallpatient numbers in the latter two groups, only the comparisonof LAA-insensitive and LAA-enhanced groups approached statistical significance (P = 0.059; Table 3). Because it appearedthat the critical factor was whether LAA had any effect at all(there was certainly no difference between patients whose cellswere suppressed versus those that were enhanced), a comparison was performed between the LAA-insensitive patients andthe LAA-sensitive (combining suppression and enhancement)patients. The MSTs of 16 and 6 months were statistically significantly different (P = 0.038).

Using the ratio definition of LAA sensitivity, the results became even more clear cut. The MST of LAA-insensitive patients (18 months) was statistically significantly different (P =0.022) from that of patients whose cell growth was enhanced byLAA (7 months). The MSTs of LAA-insensitive and LAA-suppressed patients (4 months) were not significantly different,again because of small patient number in the latter group. Asbefore, there was no difference between LAA suppression andLAA enhancement. The survival curves for LAA-insensitive,

Table 4 Survival analysis by clinical prognostic variables

eCDUl_CD

Q_

100 "|90 '

8070

60

5040

3020

10

0

LAA Effect on Colony GrowthNo Effect

â€”¿�Enhancement

Inhibition

o 12 24 36 48 60 72 84 96 108 120

Time after Bone Marrow Procurement, monthsFig. 2. Survival curves for patients whose colony growth from bone marrow

samples was insensitive to. was enhanced by. or was suppressed by addition ofLAA. LAA sensitivity is based on the ratio definition of LAA effect. A, times atwhich patients were censored.

GroupTotal

populationAge

< 50yearsAge>50yearsPrimary

MDSSecondaryMDSNo

transformation toleukemiaTransformationtoleukemiaNo

chemotherapyreceivedChemotherapyreceivedNo

chromosomeaberrationsChromosomeaberrationsFAB

1+2*FAB

3FAB 5FAB3+5FAB3+4+5Bone

marrow blasts<5%Bonemarrow blasts >5%Number

(%)519(18)42(82)40(78)11(22)24(47)16(31)36(71)15(29)8(16)16(31)14(27)22(43)

12(24)34(67)37(73)15(29)36(71)MST

G-W test,"

(months) Pvalue1314131310131313132081597

g9188T.jT.jT.jT.jT.jT.j0.490.490.580.920.0710.110.0810.0720.120.044

" G-W. Gehan-Wilcoxon.* 1, refractory anemia; 2, refractory anemia with ring sideroblasts; 3, refractory

anemia with excess of blasts: 4. chronic myelomonocytic leukemia: 5, refractoryanemia with excess of blasts in transformation.

LAA-suppressed, and LAA-enhanced patients are shown inFig. 2. Comparison of LAA-insensitive and LAA-sensitive(suppression plus enhancement) patients revealed MSTs of 18and 5 months, respectively, which were statistically significantly different (P = 0.011).

Before concluding that LAA sensitivity has prognostic value,it was necessary to examine the distribution of other prognosticfactors in the population of patients studied. A number ofunivariate analyses using the Gehan-Wilcoxon test were performed, to examine the prognostic importance of some common clinical variables (Table 4). Although there were slightdifferences in MSTs when survival was evaluated as a functionof age, whether the presenting MDS was primary or secondary,whether transformation to leukemia occurred, or whether cy-totoxic chemotherapy was received, these were not significant(P > 0.48). Although the number of patients for whom information on chromosome morphology was available was limited,there was a difference (not significant) in survival between thosepatients with a normal karyotype (MST = 20 months) andthose patients with visible chromosome aberrations (MST = 8months). There were also large differences observed when patients were grouped on the basis of FAB ranking (8), but thesewere still not significant.

A statistically significant difference was approached (P =0.072) when patients were grouped on the basis of FAB category, dichotomized as 1+2 versus 3+5. Since FAB categoriesare based in large part on percentage of bone marrow blasts (8),it is not surprising that a similar results were obtained for FABrank and for percentage of bone marrow blasts. For percentageof bone marrow blasts (dichotomized as <5% or >5%) therewas a significant (P = 0.044) difference in MST (18 versus 8months). Finally, it is critical to note that there was no changein survival duration over the course of this study period, i.e.,patients entered in the first 5 years did not exhibit a differentsurvival than did those entered in the second 5 years.

In order to dismiss the possibility that percentage of bonemarrow blasts and the ability to form colonies or exhibit an

4461




LAA effect are functionally correlated, an analysis was performed of the distribution of percentage of bone marrow blastsin three groups, as defined by LAA ratio. The percentages ofbone marrow blasts were 8.2 Â±6.5, 11.2 Â±8.6, and 12.9 Â±9.4(mean Â±SD), respectively, for patients in the no-growth, LAA-insensitive, and LAA-sensitive groups. These differences werenot significant (P > 0.10), as demonstrated using a Student-Newman-Kuels test. If the alternate, t test, definition of LAAsensitivity was used, the distributions were again not different:8.2 Â±6.5, 10.9 Â±8.7, and 13.7 Â±9.3, respectively. Thus, therewas no apparent correlation between percentage of bone marrow blasts and colony growth or LAA effect.

Returning to a conventional survival curve format, the survival patterns for patients in the LAA-sensitive range (as defined by LAA ratio) and for those in the LAA-insensitive rangeare plotted in Fig. 3A. Likewise, survival is plotted as a functionof the known prognostic variable, percentage of bone marrowblasts (Fig. 3B). As indicated in Table 4, there was a significantdifference (P = 0.044) between the MSTs of 18 and 8 monthsobtained for patients with <5% blasts or >5% blasts, respectively. Thus, it would appear that LAA sensitivity is as valuablea prognostic predictor as is percentage of bone marrow blasts.

If we now take into consideration both LAA sensitivity andpercentage of bone marrow blasts, four prognostic subgroupscan be analyzed (Fig. 4; Table 5). For LAA-sensitive (defined byLAA ratio) patients, percentage of bone marrow blasts remained a significant (P = 0.0059) prognostic variable (MSTs of23 versus 4 months). For LAA-insensitive patients, percentageof bone marrow blasts differentiated patient survival further.

ALAA Effect on Colony Growth

â€”¿�LAA-insensitive

â€”¿�LAAâ€”Sensitive

12 24 36 48 72 84 96 108 120

100 BBone Marrow Percent Blasts

ID(/)-Ã•-Â»C(UuL_CDCL70605040302010n.1

L - vo*â€¢

\ L. - - -5-30^â€¢\ L-n1Ui_1

Al

\""-â€¢*--_*

0 12 24 36 48 60 72 84 96 108 120

Time after Bone Marrow Procurement, monthsFig. 3. A, survival curves for patients whose bone marrow samples were insen

sitive to LAA growth modulation and those that were sensitive to LAA modulation. The curves were constructed from the data in Fig. 2. B, survival curves forpatients dichotomized to <5% or >5% bone marrow blasts. A, times at whichpatients were censored.

_>(7>-4->Ol_D_100'

90

80

706040

302010

nA.....A

â€”¿�.LAA Effect and Bone Marrow Blasts

â€¢¿�',"! No LAAEffect/BM<5*I

No LAAEffect/BM5-30*LAAEffect/BM <5%

---â€¢LAAEffect/BM5-30*i.'

::'--: A i

0 12 24 36 48 60 72 84 96 108 120

Time after Bone Marrow Procurement, monthsFig. 4. Survival curves for four prognostic groups denned by LAA sensitivity

(LAA ratio) and percentage of bone marrow blasts (BM) (<5% or >5%). A, timesat which patients were censored.

with marginal statistical significance (P = 0.044; MSTs of >23

and 16 months). Viewed conversely, for the group of patientswith bone marrow blasts of >5Â°/o,LAA sensitivity providedadditional statistically significant (P = 0.0063) prognosticinformation: MSTs of 16 versus 4 months for LAA-insensitiveand LAA-sensitive patients, respectively. There was no differ

ence, based on LAA sensitivity, for patients with bone marrowblasts of <5% (MSTs of 23 months). Similar results were obtained when the t test definition of LAA sensitivity was employed.

Next, a multivariate analysis was performed to determinewhether LAA sensitivity and percentage of bone marrow blastswere, in fact, truly independent prognostic variables. The multivariate Cox proportional hazard model was used, with survival time as the dependent variable and a number of clinicaland clonogenic variables as the covariates to be tested. Thevariables evaluated included age, primary or secondary development of MDS, the presence or absence of chromosome aberrations, bone marrow blasts, chemotherapy received or not,transformation to leukemia, and LAA effect. At first, FABcategory was also entered, with a separation into ranks 1+2versus 3+4+5. However, since this dichotomization is in parallel with the dichotomization of percentage of bone marrowblasts, FAB category was not considered as a separate prognostic variable.

Initially, the analyses were run with percentage of bone marrow blasts and age entered as either continuous variables ordichotomized (at 5% and age 50, respectively). Since both approaches yielded essentially identical results, the variables wereentered as dichotomous variables for subsequent analyses. In asimilar fashion, LAA effect was first categorized into threepossibilities: no cell growth, no LAA effect, and LAA sensitivity. The results indicated that only LAA sensitivity (P = 0.024)and bone marrow blasts of >5% (P = 0.033) were independentpredictors of decreased survival. When the number of LAAcategories was expanded to denote whether enhancement orinhibition was the specific effect of LAA, the same predictors ofsurvival were identified (P = 0.028 and P = 0.022, respectively).

In order to verify the findings described above, the Cox proportional hazard model was run in a non-stepwise fashion, using only the variables that had been identified as at least marginally (P < 0.10) significant. The results of this analysisconfirmed that both percentage of bone marrow blasts and LAAmodification of colony formation were significant predictors ofsurvival outcome (P < 0.05).

4462



GROWTH MODULATION BY t-ASCORBIC ACID

Table 5 Survival analysis by LAA sensitivity and bone marrow blasts

GroupLAAeffect"LAA-insensitive,

BM* <5%LAA-insensitive,BM >5%LAA-sensitive.

BM <5%LAA-sensitive,BM >5%LAA

effect'LAA-insensitive,BM <5%LAA-insensitive,BM >5%LAA-sensitive,

BM <5%LAA-sensitive,BM >5%Prognostic

signs (LAA effect defined by ttest)NoriskfactorsOne

risk factorTwo riskfactorsPrognostic

signs (LAA effect defined byratio)NoriskfactorsOnerisk factor

Two riskfactors"Defined by P of <0.05 for individual itests.*BM, bone marrowblasts.cDefined by ratio of number of colonies with or withoutNumber7163II614413719

11618

13LAA.MST

(months)>23

-|11

h23\-4>23

-,16\-23

[-4L>23

-iâ€¢J1>23

-,".10.0160.270.00220.0440.220.00470.016

0.00220.063

0.0047P

value1h

0.47 j-0.11L0.088-|\-

0.37 T-0.0059L0.00631

0.0525L

0.0016

Having demonstrated that bone marrow blasts of >5% andcolony growth modification by addition of LAA are both independent predictors of increased hazard (i.e., poor survival prognosis), we categorized patients into three groups, dependingupon whether the patient exhibited no risk factors (low bonemarrow blasts and LAA insensitive), one risk factor (either lowblasts and LAA sensitive or high blasts and LAA insensitive), ortwo risk factors (both high blasts and LAA sensitive). Analyzedin this fashion (Fig. 5; Table 5), a statistically significant difference was observed between the two-risk factor group andboth the zero- and the one-risk factor groups (P = 0.0047 and0.0016, respectively). Further, the separation between the one-and the zero-risk factor groups approached statistical significance (P = 0.063). The same results were obtained when the itest definition of LAA sensitivity was used, except that thedifference between zero and one risk factors was more obvious(P =0.016).

DISCUSSION

In this study, a clear difference in colony growth pattern wasshown between normal bone marrow and bone marrow ofMDS. On the other hand, the growth pattern of MDS isremarkably close to that of AML, in terms of both the numberof colonies formed and the effect of daily feeding on colonygrowth (5, 9). Specifically, the effect of LAA reported for AML(5), where 33% (of 163 cases studied) exhibited growth enhancement by LAA and 17% exhibited growth suppression, issimilar to that shown here for MDS (24% enhancement and14% suppression). A few normal cases demonstrated a modestdegree of suppression by LAA, but none of the normal samplesshowed enhancement. Moreover, it has been shown that thisLAA effect is a true biological effect and not a simple physico-chemical effect (e.g., pH), since the optical isomers of LAA donot modulate colony growth (5). The similarity between AMLand MDS is unlikely to be a coincidence; 34 cases of chronicmyelogenous leukemia similarly studied in our laboratoryshowed a pattern similar to that of normal bone marrow butdissimilar from that of MDS or AML.

One might speculate that growth modulation by LAA is related to the requirement for daily feeding with medium that is

>'>Percent

Sur90807060

50

40

30

2010

nSurvive

,.\;:

'"k.

i _( Zero. ! | Zero. â€¢¿�.'<One\iby

PrognosticRiskZero

Risk Factors-â€¢OneRisk Factor--Two RiskFactorsAvs

Two Risk Factorsvs One Risk Factor's Two RiskFactors'

AFactors

NMST)ZO18

16

134P=0.0047

P=0.063P=0.0016

0 12 24 36 48 60 72 84 96 108 120

Time after Bone Marrow Procurement, monthsFig. 5. Survival curves for three groups of patients defined by the presence of

prognostic risk factors, based on LAA sensitivity and percentage of bone marrowblasts (<5% or >5%). Patients were categorized as having zero risk factors (i.e.,low bone marrow blasts and LAA insensitivity), one risk factor (either low blastsand LAA sensitivity or high blasts and LAA insensitivity), or two risk factors (i.e.,high blasts and LAA sensitivity). A, times at which patients were censored.

exhibited by both MDS and AML cells. Having identified theneed for one growth factor in conditioned medium, it is reasonable to expect that there may be further modulation by LAA (orother growth factors). As pointed out, the daily feeding effectitself was not of prognostic value for survival. However, it wasobserved that the patients that displayed the greater effect ofdaily feeding also tended to display LAA sensitivity. Only fourof 17 that had a <3-fold increase with daily feeding were alsoLAA sensitive, while 13 of 20 with a >3-fold increase were LAAsensitive. This difference was statistically significant (P =0.012) by x2 contingency table analysis. Even more striking, of

the 11 patients that showed LAA enhancement, 10 of these hadcolony number increased >3-fold by daily feeding. When thethree categories of daily feeding effect (<3-fold, >3-fold, andabsolute requirement) were cross-factored with the three categories of LAA response (suppression, no effect, and enhancement), there was a definite correlation (Table 6, P = 0.0044).

4463




Table 6 Contingency test for feeding effect versus LAA effect

Dailyfeedingratio<3-fold

>3-foldAbsoluteTotalLAA

effect"Suppression3

036No

effect13

6120Enhancement173

11Total17

137

371x2 with 4 degrees of freedom, 15.17; probability level, 0.0044.

Thus, while the response to daily feeding and the growth-mod-ulatory effects of LAA were related, it was only the LAA effectthat was shown to be a critical variable in terms of patientsurvival (Table 3).

It is interesting that the addition of LAA had two quite different effects on colony formation: suppression or enhancement. However, these were not chance effects; whether therewas enhancement or suppression was consistent for individualpatients. This has been substantiated by multiple repeat assaysfrom the same bone marrow aspirations. In fact, repeat samplesfrom the same patient, even when the second sample was obtained subsequent to treatment, demonstrated the same qualitative (whether suppression or enhancement) effect of LAAaddition. One could postulate that LAA plays a role in somecritical metabolic pathway involving oxidation-reduction reactions or electron transfer, which might finally lead to either freeradical scavenging or free radical production. While alterationof free radical levels could conceivably result in either suppression or enhancement of colony formation, there is no directevidence that this is the case. Thus, at this time, no definiteexplanation is available for why colony-forming cells of MDSor AM L from one patient are enhanced while cells from another patient are suppressed. But, regardless of the exact mechanism for the LAA effect in vitro, in terms of prognostic valueit is only important that LAA modulate colony growth; it isnot critical whether there is enhancement or suppression ofgrowth.

The predictive value of the LAA effect for patient prognosismay be a function of the general tendency of cells to have theirgrowth patterns modified and not something specifically related to LAA. Cells that are inherently susceptible (in vitro) togrowth modulation might also express a similar growth controlin situ. One could conjecture that if a cell population was amenable to growth control then it might be affected by fluctuatinglevels of growth factors (LAA or other, nonspecified, factors) inserum. According to this rationale, the direction of the modulation (suppression versus enhancement) would be less important than the fact that any modification of growth is possible. Itshould be pointed out that there was essentially no differencebetween the 75% of the patients that formed colonies and the25% that did not, either in terms of distribution of prognosticfactors or in terms of survival. Nor did multivariate analysisidentify the ability to form colonies as a significant predictor ofsurvival.

A reliable and highly sensitive prognostic factor, such as LAAeffect, can be very useful in making clinical decisions for management in diseases such as MDS, which has a highly variableprognosis with no uniform or easy treatment (14-16). SinceLAA sensitivity is quantitative as well as objective, it should beparticularly useful in the precise and unbiased analysis of prognostic factors. In this study, the cell culture test for LAA effectwas proven to be a more powerful prognostic factor than percentage of bone marrow blasts. Importantly, there was no significant correlation between LAA sensitivity and percentage of

bone marrow blasts. Therefore, both these factors can be usedto identify extremely favorable and poor prognosis groups (Fig.5). This distinction may assist therapeutic decision-making involving high risk treatment, e.g., bone marrow ablation followed by allogeneic transplant (15, 16).

One important question remaining is whether this in vitrofinding of suppression of abnormal clones by LAA deprivation(or LAA supplementation, in a minority of patients) can bereproduced in vivo. Using this culture system and another, similar, system, it has been shown that there is a good correlationbetween in vitro cytotoxicity of chemotherapeutic drugs in malignant cells obtained from patients and the clinical response ofthe same patients to the same drugs (17-20). In part, this isbecause the cells used in this system are obtained freshly frompatients (rather than using cell lines or animal cells) for directclinical correlation with an individual patient response.

Increasing evidence indicates that colony-stimulating factors,i.e., growth factors for hemopoietic cells commonly studiedwith an in vitro "colony assay" similar to ours, have in vivo

relevance in humans, with colony-stimulating factors for leu

kocytes being used in various leukopenic conditions (21). It haslong been recognized that tumor cells, especially leukemic cells,are amenable to biological growth modulation, such as withcolony-stimulating factors (21, 22), and this is beginning to betested in clinical trials (21, 23-25). There are in vivo studiesshowing that, in guinea pig leukemia (26) and a solid tumor(27), growth is suppressed by LAA depletion. Therefore, it isconceivable that the abnormal clones that are sensitive to LAAdepletion in vitro may also be suppressed in vivo by LAA deficiency.

The current study, documenting the clinical implication ofthe LAA effect, gives further impetus to the possibility of usingthis effect in a therapeutic fashion. It is feasible to test thispossibility, because the LAA plasma concentration can be reduced to a near-zero level in 1 month (28, 29) and maintainedat this level for 3 to 5 months, in humans on scorbutic diets,before any significant clinical evidence of scurvy develops (30,31). A trial of LAA supplementation in LAA-suppressed cases

would also be possible. LAA supplementation is simple andsafe, as shown in a trial on colon cancer (32), but no controlledtrial of LAA supplementation has been conducted in other malignancies. It would be much easier to conduct this clinical trialin MDS than in AML; MDS is quite often a slow diseaseprocess, and there is no standard treatment available. Moreover, the prognosis is dismal, with a median survival of <1 yearfor certain subsets (14) and especially, as shown in this study,for LAA-modulated cases, which would be prime candidates foran LAA clinical manipulation trial. A note of caution is thatstratification is critical, especially between the opposing LAA-enhanced and -suppressed populations, in the event that a clinical trial involving LAA is contemplated. A simple randomization without this stratification might result in no overall effect,because the benefit obtained by one population might be offsetby a detrimental effect in the opposing population.

It may be too simplistic to expect a therapeutic benefit fromsimple in vivo manipulation of LAA level. Even if this straightforward LAA manipulation approach fails, there remains theexistence of differences in some vital metabolic pathway(s) between certain MDS and normal bone marrow cells. The elucidation of the mechanism(s) for these differences may eventuallylead to a more sophisticated and effective approach to the management of patients with MDS.

4464



APPENDIX


LAA effect in vitro on colonies of MDS (repeat cultures for consistency checks)

No. ofcolonies''Expt."'*AlA2BlClC2DlElE2E3E4FlGlMDS

caseno.c6Normal67NormalAL-146*15Normal15/2''18Normal21AL-263J21AL-242*21/2'21/2'3033Normal3840NormalCellno./dish5x10s5xIO51

xIO55xIO55xIO55x

IO52xIO55xIO55xIO55x10sSx10s5xIO55x10s5xIO55xIO45

xIO52xIO55x10s3xIO45

xIO55x10"5

xIO55XIO5Without

LAA807

Â±6450Â±747

Â±527Â±7196

Â±129.8Â±0.30110

Â±5745Â±8576

Â±2177+16045

Â±52.1Â±1.5000.7

0.319851621471485

7.66282143

4.81076.8With

LAA1547

Â±21541+5320

104.721221049

1710219124

161590391261099

Â±14110Â±1045

Â±3186Â±3343

Â±319.5Â±428

Â±122998Â±95146

Â±5.688Â±5.4792Â±74116+3.7109Â±8.6t

test'snvhsShssvhsnvhshsvhsvhsnvhsvhsn"svhsnnnvhsnRatioÂ±LAA/1.920.826.880.180.625.00QO1.132.131.660.56OO1.0089.0008042.01.511.001.031.262.691.01Ratio+ LAA

of initialculture*2.04,

E2.04,

E0.26,S104,

E104,

E2.08,ECÂ«OC,

fc.Â»,

Eoc,E7.67,

E0.93,N0.92,

N2.15,E

" Attempts were made to repeat cultures on all LAA-modulated cases if cells were available. There were multiple repeat cultures on LAA-insensitive cases withconsistent results, but they are not shown in this table.

* Al and A2, first and second repeat experiments for the same case.c Numbers without prefix are MDS case numbers; normal indicates normal bone marrow; numbers with prefix of AL are acute myeloid leukemia case numbers.d Number of colonies expressed as mean Â±SE.' t test for the significant difference between numbers of colonies with and without LAA: s. significant (P < 0.05); hs, highly significant (P < 0.01 ); vhs, very highly

significant (P < 0.001); n, not significant (P> 0.05).^Ratio of numbers of colonies with (+)/without(â€”) LAA. Â»,absolute requirement for growth.* For comparison, initial culture results are shown in terms of the ratio with and without LAA and the initial designation as to whether LAA enhances (E), suppresses

(S), or does not affect (N) colony-forming ability.* A leukemic patient well established to have LAA enhancement by seven repeated experiments over 3 years (from May 1, 1981, to August 24, 1984). The ratios of

number of colonies with and without LAA are 680, Â«,5.0 (shown in this table), 79, 22, 40, 62, and 87.1A second bone marrow (15/2) was obtained 9 months after the initial marrow (15). This had 20% blasts, but 15/2 had 90% blasts, indicating leukemic transition (M5

type by FAB classification).J An AL case known to be LAA independent, serving as control.* Another leukemic patient well established to have LAA enhancement by eight repeated experiments over a 4-year period (from October 13,1982, to October 1,1986).

The ratios of number of colonies with and without LAA are 27.3, 68.6, *, * (shown in this table), 114, 50, *, and 117.'A second bone marrow (21/2) was obtained 6 months after the initial marrow (21). This had 5% blasts and 21/2 had 20%.m Only three dishes were not lost due to bacterial contamination. Therefore, the power (1 - ÃŸerror) of the t test was too low to detect significance.

ACKNOWLEDGMENTS

We thank all contributing hematologists/oncologists and patholo-gists for providing bone marrow samples and clinical/hematologicalinformation on patients with myelodysplastic syndromes.

REFERENCES

1. Park, C. H., Bergsagel, D. E., and McCulloch, E. A. Ascorbic acid: a culturerequirement for colony formation by mouse plasmacytoma cells. Science(Washington DC), 174: 720-722, 1971.

2. Park, C. H. Studies of growth characteristics of myeloma in vitro. Ph. D.dissertation, pp. 52-62. Toronto: University of Toronto, 1971,

3. Park. C. H., Amare, M., and Hoogstraten, B. Analysis of the growth enhancing effect of L-ascorbic acid on human leukemic cells in culture. Exp. Hema-tol.. 8: 853-859, 1980.

4. Park, C. H., Amare, M., Savin, M. A., and Hoogstraten, B. Growth suppression of human leukemic cells in vitro by L-ascorbic acid. Cancer Res., 40:1062-1065, 1980.

5. Park, C. H. Biological nature of the effect of ascorbic acids on the growth ofhuman leukemic cells. Cancer Res., 45: 3969-3973, 1985.

6. Park, C. H., and Kimler. B. F. Growth modulation of human leukemic,preleukemic, and myeloma progenitor cells by L-ascorbic acid. Am. J. Clin.Nutr., 54: 1241S-1246S, 1991.

7. Alcain, F. J., Buron, M. I., Rodriguez-Aguilera, J. C., Villalba, J. M., andNavas, P. Ascorbate free radical stimulates the growth of a human promye-locytic leukemia cell line. Cancer Res., 50: 5887-5891, 1990.

8. Bennett, J. M., Catovsky, D., Daniel, M. L, Flandrin, G., Galton, D. A. G.,Gralnick. H. R., and Sultan, C. The French-American-British (FAB) Cooperative Group. Proposals for the classification of the myelodysplastic syndromes. Br. J. Haematol., 51: 189-199, 1980.

9. Park. C. H., Savin, M. A., Hoogstraten, B., Amare. M., and Hathaway, P.

Improved growth of in vitro colonies in human acute leukemia with thefeeding culture method. Cancer Res., 37: 4595-4601, 1977.

10. Feng, J., Melcher, A. H., Brunette, D. M., and Moe, H. K. Determination ofL-ascorbic acid levels in culture medium, concentrations in commercial mediaand maintenance of levels under conditions of organ culture. In Vitro, 13:91-99, 1977.

11. Mohberg, J., and Johnson, M. J. Stability of vitamins in a chemically definedmedium for 929-L fibroblasts. J. Nati. Cancer Inst., 31: 603-610, 1963.

12. Park, C. H. Cell-cycle manipulation of human leukemic progenitor cells withhumoral adjustment in vitro. Blood, 58: 179-182, 1980.

13. Metcalf, D. Outline of hemopoiesis and current terminology. In: Hemopoi-etic Colonies, pp. 5-11. New York: Springer-Verlag, 1977.

14. Koeffler, H. P. Myelodysplastic syndromes (preleukemia). Semin. Hematol.,25:284-299, 1986.

15. Cheson, B. D. The myelodysplastic syndromes: current approaches to therapy. Ann. Intern. Med., 112: 932-941, 1990.

16. Appelbaum. F. R., Barrali, J., Storb, R., Fisher, L. D., Schoch, G., Ramberg,R. E., Slmili..in, H., Anasetti, C., Bearman, S. I., Beatty, P., Bensinger, W. I.,Buckner, C. D., Clift, R. A., Hansen, J. A., Martin, P., Petersen, F. B.,Sanders, J. E., Singer, J., Stewart, P., Sullivan, K. M., Witherspoon, R. P.,and Thomas, E. D. Bone marrow transplantation for patients with myelo-dysplasia. Pretreatment variables and outcome. Ann. Intern. Med., 112:590-597, 1990.

17. Park, C. H., Amare, M., Savin, M. A., Goodwin, J. W., Newcomb, M. M.,and Hoogstraten, B. Prediction of chemotherapy response in human leukemia using an in vitro chemotherapy sensitivity test on the leukemic colony-forming cells. Blood, 55: 595-601, 1980.

18. Park, C. H., Wiernik, P. H., Morrison, F. S., Amare, M., Van Sloten, K., andMaloney, T. R. Clinical correlations of leukemic clonogenic cell chemosen-sitivity assessed by in vitro continuous exposure to drugs. Cancer Res., 43:2346-2349, 1983.

19. Salmon, S. E., Hamburger, A. W., Soehnlen, B., Durie, B. G. M., Alberts, D.S., and Moon, T. E. Quantitation of differential sensitivity of human tumor

4465



GROWTH MODULATION BY u-ASCORBIC ACID

stem cells to anticancer drugs. N. Engl. J. Med., 298: 1321-1327, 1978.20. Von Hoff, D. D., Casper, J., Bradley, E., Sandbach, J., Jones, D., and

Makuch, R. Association between human tumor colony-forming assay resultsand response of an individual patient's tumor to chemotherapy. Am. J. Med.,70: 1027-1032, 1981.

21. Groopman, J. E., Molina, J. M., and Scadden, D. T. Hematopoietic growthfactors. Biology and clinical applications. N. Engl. J. Med., 321: 1449-1459,1989.

22. Sachs, L. Control of normal cell differentiation and the phenotypic reversion of malignancy in myeloid leukemia. Nature (Lond.), 274: 535-539,1979.

23. Muhm, M., Andreeff, M., Geissler, K., Gorischek, C, Haas, O., Haimi, J.,Hinterberger, W., Schulz, G., Speiser, W., Sunder-Plassmann, G., Valent, P.,Lechner, K., and Bettelheim, P. RhGM-CSF in combination with chemotherapy: a new strategy in the therapy of acute myeloid leukemia (Abstract).Blood, 74 (Suppl. 1): 117A, 1989.

24. Hoelzer, D., Ganser, A., Seipelt, G., Ottmann, O., Hoffken, K., Boogaerts,M., Verhaef, G., Lutz, D., Krieger, O., de Witte, T., Diehl, V., Lathan, B.,Ferrant, A., Martial, P., Klausmann, M., Gangji, D., Frisch, J., and Schulz,G. Simultaneous treatment with recombinant human granulocyte-macroph-age colony-stimulating factor (rhGM-CSF) and low-dose cytosine arabino-side (LD-ara-C) in patients with myelodysplastic syndrome (Abstract). Blood,

74 (Suppl. 1): 118A, 1987.25. Estey, E. H., Dixon, D., Kantarjian, H. M., Keating, M. J., McCredie, K.,

Bodey, G. P., Kurzrock, R., Talpaz, M., Freireich, E. J., Deisseroth, A. B.,and Gutterman, J. U. Treatment of poor-prognosis, newly diagnosed acutemyeloid leukemia with ara-C and recombinant human granulocyte-macroph-age colony-stimulating factor. Blood, 75: 1766-1769, 1990.

26. Nadel, E. M. History and further observations (1954-1976) of the L2Cleukemia in the guinea pig. Fed. Proc., 36: 2249-2254, 1977.

27. Migliozzi, J. A. Effect of ascorbic acid on tumor growth. Br. J. Cancer, 35:448-453, 1977.

28. Jacob, R. A., Skala, J. H., and Omaye, S. P. Biochemical indices of humanvitamin C status. Am. J. Clin. Nutr., 46: 818-826, 1987.

29. Blanchard, J. Depletion and repletion kinetics of vitamin C in humans. J.Nutr., 121: 170-176, 1991.

30. Crandon, J. H., Lund, C. C., and Dill, D. B. Experimental human scurvy. N.Engl. J. Med., 223: 353-369, 1940.

31. Peters, R. A., for the Accessory Factors Committee. Vitamin requirement ofhuman adults. Lancet, /: 853-858, 1948.

32. Moertel, C. G., Fleming, T. R., Creagan, E. T., Rubin, J., O'Connell, M. J.,and Ames, M. M. High-dose vitamin C versus placebo in the treatment ofpatients with advanced cancer who have had no prior chemotherapy. N. Engl.J. Med., 312: 137-141, 1985.

4466



1992;52:4458-4466. Cancer Res Chan H. Park, Bruce F. Kimler, David Bodensteiner, et al. Myelodysplastic SyndromesColony-forming Cells from Bone Marrow of Patients with

Growth Modulation by l-Ascorbic Acid ofIn Vitro

Updated version

http://cancerres.aacrjournals.org/content/52/16/4458

Access the most recent version of this article at:

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

Subscriptions

Reprints and

[email protected] at

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

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/52/16/4458To request permission to re-use all or part of this article, use this link



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

mailto:[email protected]



in vitro growth modulation by l-ascorbic acid of colony ... · 51 patients with mds, received from...

Documents