| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical Investigations |
Department of Hematology and Genetic Pathology, Flinders University of South Australia and Flinders Medical Center, South Australia 5042 [M. J. B., P. J. S., G. D., E. H., S-H. N., L. E. S., A. A. M.]; Department of Laboratory Medicine, West China University of Medical Sciences, Chengdu 610041, People’s Republic of China [L. P.]; and Department of Hematology/Oncology, Women’s and Children’s Hospital, North Adelaide, South Australia 5006 [I. R. G. T., M. S. R.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The early-mutation hypothesis predicts that the leukemic population at diagnosis will be heterogeneous and contain one or more subpopulations of resistant cells, whereas the late-mutation hypothesis predicts that the leukemic population at diagnosis will be relatively homogeneous and that any resistant subpopulations will arise later. Drug resistance in vivo can be measured by the rate of decline of the leukemic population during drug treatment. We therefore used molecular techniques to study the decline of MRD during induction therapy to examine these two hypotheses.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Tissue Samples and DNA.
Bone marrow aspirates and peripheral blood samples were obtained at
diagnosis, and bone marrow, aspirates, trephines, and peripheral blood
samples were obtained on days 14, 21, and 35 of induction treatment.
Samples were not obtained from some patients on both days 14 and 21.
Cells were removed from trephines by crushing the sample and
continuously washing with saline containing 2.5 µM
K2 EDTA, as described previously
(1)
. DNA was extracted by the standard methods of
proteinase K digestion, followed by phenol-chloroform extraction and
ethanol precipitation.
Treatment.
Systemic induction therapy consisted of daily oral prednisolone
(40 mg/m2) for 4 weeks with tailing off over the
fifth week; vincristine (1.5 mg/m2) and
daunorubicin (25 mg/m2) weekly x 4 on the first day of each week; and erwinia L-asparaginase
(6000 units/m2) three times weekly for 3 weeks.
The final doses of vincristine, daunorubicin, and asparaginase were
given on day 21.
Quantification of Leukemia.
Leukemia was quantified by using the CDR3 region of the rearranged
immunoglobulin heavy chain (IgH) gene as a molecular marker for cells
of the leukemic clone (2)
and the N-ras gene as
a molecular marker for all cells. For each patient the rearranged IgH
gene was sequenced, and a pair of "leukemia-specific" primers was
synthesized. The numbers of amplifiable leukemic targets and
amplifiable N-ras targets were quantified by PCR using
limiting dilution analysis with Poisson statistics (3
, 4)
,
and the proportion of nucleated marrow cells that belong to the
leukemic clone was calculated assuming each leukemic cell contains one
rearranged IgH gene and all cells contain 2 N-ras genes. For
samples in which leukemia could not be detected, an upper limit value
for the level of MRD could be calculated from the number of amplifiable
genomes studied.
Marrow.
Leukemia could be quantified in 27 of the 40 patients (67.5%), the
main reason for failure being technical difficulties in obtaining an
IgH sequence for use as a clonal marker. Five of the 27 patients had
marrow samples only at diagnosis and at the end of induction treatment
and were excluded from this study because their results could provide
no information on serial changes in MRD. Their response to induction
appeared comparable with those of the other patients. Analysis of MRD
was based on the remaining 22 patients who provided marrow samples on
57 occasions. Leukemia was detected and quantified on 51 of these
(89%), and on the remaining six occasions an upper limit value for the
level of MRD was estimated.
Some estimations of MRD in individual patients had been performed either in duplicate on the same sample or by studying a trephine specimen in addition to an aspirate. Duplicate estimations were pooled by calculating the geometric mean of both results or, if MRD had not been detected in one or both estimations, by pooling the raw data and calculating the actual value for MRD or the upper limit value. Because most published MRD values are based on aspirates, trephine estimates of MRD were converted to an aspirate value by dividing the level of MRD by 4.1. This factor, which has been determined previously as representing the mean ratio between trephine and aspirate MRD levels (1) , probably results from lesser contamination of trephines by peripheral blood. Pooling with actual aspirate MRD levels was then performed as described above for replicate estimations on the one sample.
Blood.
Levels of leukemia at diagnosis were calculated from the white cell
count and the differential count, as described previously
(5)
. Fifteen of the 22 patients provided blood samples for
study, but 2 of them never had leukemia detected in their blood and
were therefore excluded from this part of the study. The remaining 13
patients provided a total of 31 blood samples after diagnosis. Leukemia
was detected and quantified in 19 of these (61%), and an upper limit
value estimated for the remainder.
Calculation of Absolute Levels of MRD.
The cellularity of both marrow and blood is known to vary substantially
during induction treatment. Because of this variation, the use of
relative values for MRD, i.e., leukemic cells/total cells,
is misleading because it will lead to a systematic underestimation of
the rate of decline of leukemia during periods when total cellularity
is decreasing and a systematic overestimation during periods when total
cellularity is increasing. The data for relative MRD levels were
therefore converted to an absolute number of leukemic cells/unit volume
by multiplying the relative MRD level and the absolute number of
cells/unit volume. For blood, this was straightforward because the
absolute number of leukocytes/liter was known. For marrow, cells/unit
volume were estimated by counting the number of cells recovered per
trephine and assuming that the mean trephine volumes at each stage of
treatment remained the same. This assumption would not have biased any
estimates of means, although it would have increased the variance. The
logarithmic means of cellularities were 2.9 x 105 cells at day 14 (n = 18), 1.0 x 105 at day 21
(n = 16), and 6.9 x 105 at day 35 (n = 21;
Ref. 1
). An additional 13 marrows, reported as being of
normal cellularity, were obtained from patients 1–2 years after
diagnosis, and the logarithmic mean of cells recovered from the
trephines was 3.2 x 106. No
marrow trephines were available at diagnosis. Because the marrow at
diagnosis is usually judged to be "packed" with leukemic cells, we
assumed that the cellularity at diagnosis was 2 x normal. We also assumed that the mean cellularity of the year 1–2
marrows was 1 x normal. Both of these assumptions,
almost certainly underestimated the true differences from normal
values; if so, they were conservative, tending to act against the
detection of significant differences.
Calculation of the Rate of Decline of the Leukemic Clone during
Early and Late Induction.
The rate at which the leukemic clone declined was expressed as the
fraction surviving per week (FS/W) during treatment. This
value is given by: FS/W = W
,
where N1 and
N2 are the numbers of leukemic cells
at the beginning and end of a time of w weeks.
N1 and
N2 can be expressed as absolute
numbers or as fractions of day 0 values. It should be noted that the
actual volume of marrow or blood used for estimation of cells/unit
volume does not affect the final calculation, because it is present in
both numerator and denominator and cancels out.
Actual or limit values for FS/W were calculated between day 0 and day 14, between day 14 and day 21, and between day 21 and day 35 by assuming a constant exponential rate of decline between the two time points of interest and calculating the FS/W either from two actual values or from one actual value and one upper limit value.
Statistics.
The significance of observed differences in FS/W for
different time intervals were tested using the two-tailed Mann-Whitney
U test, unpaired except where indicated. For some samples,
particularly blood, only an upper limit value could be estimated for
the level of MRD, and thus only an upper or lower limit value could be
calculated for FS/W. For such samples the true value of MRD
lies somewhere between the upper limit value and zero. Discarding such
samples or equating the true value to the upper limit value would
introduce a systematic upward bias, whereas equating the true value to
zero would introduce a systematic downward bias. For this reason, we
made assumptions as to the true MRD values for samples with upper limit
values and performed a sensitivity analysis to test the effects of
varying these assumptions. If an upper limit value was followed by a
later actual MRD value that was greater than that upper limit value,
suggesting that the leukemic population had not declined between the
two time points, we regarded the level of MRD at the earlier time point
as being 0.5 x the upper limit value; otherwise we
regarded the level of MRD as being 0.1 x the upper
limit value.
Sensitivity Analysis.
The assumptions made with regard to marrow cellularity and upper limit
values would affect the significance values obtained from testing
observed differences. A sensitivity analysis was therefore performed by
varying the assumptions to observe the effect on significance values
and on conclusions drawn from them. The assumed relationship to normal
cellularity for day 0 and year 1–2 marrows affects only the
FS/W calculated for days 0–14, because the FS/W
calculated for the other time intervals is based on actual
measurements. Significance values were therefore also calculated,
assuming that the cellularity on day 0 was equal to the cellularity of
normal marrow. For the transformations of upper limit values, the
assumed values of MRD were varied upwards or downwards within wide
limits, FS/W was calculated, and significance testing was
performed.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
shows, for each patient, the level of leukemia in marrow and blood at
various times during induction. The relative MRD levels for marrow were
first converted to absolute numbers of leukemic cells per unit of
volume and then normalized by dividing by the day 0 numbers; the values
for blood are absolute cell numbers/liter. The means ± 1 SE on different days for the normalized levels of MRD in marrow and
blood are shown in Fig. 1
. For days 14, 21, and 35, the respective values were: for marrow,
5.2 x 10-5 (± 1 SE,
9.9 x 10-5 to 2.7 x 10-5), 2.2 x 10-6 (4.9 x 10-6 to 9.5 x 10-7), and 9.9 x 10-6 (2.1 x 10-5 to 4.8 x 10-6); and for blood, 1.7 x 10-5 (1.1 x 10-4 to 2.5 x 10-6), 5.7 x 10-6 (2.0 x 10-5 to 1.6 x 10-6), and 7.4 x 10-6 (3.7 x 10-5 to 1.5 x 10-6). The results indicate a rapid decline in
leukemia during the first 2 weeks of treatment, a slowing of decline
during the third week, and a plateau during the fourth and fifth weeks.
The rates of decline in individual patients are shown in Fig. 2
, which shows the FS/W of the leukemic clone in marrow and
blood between days 0 and 14, days 14 and 21, and days 21 and 35.
Because of the missing samples, in several patients FS/W
could not be determined for these intervals, but data were available
for days 0–21 or days 14–35, and these are also shown.
|
|
|
The sensitivity analysis confirmed the reality of the differences
between the FS/W for the different time intervals, because
significant differences persisted, despite wide variations in the
magnitude of the assumptions used. For marrow, the differences between
the results for the three time intervals remained significant when
cellularity on day 0 or day 35 was equated to the cellularity of normal
marrow (P between 0.02 and 0.001); when all assumed MRD
values were equated to the observed upper limit values (Ps
between 0.01 and 0.0003); and when the assumed values for marrow MRD
were decreased to any extent, even to zero (Ps between 0.01
and 0.0003). For blood, the differences between the various time
intervals remained significant until assumed MRD values were decreased
by a further two orders of magnitude, at which point only the
difference between day 0–14 and day 21–35 values remained significant
(P = 0.04). Thus, the results for both marrow
and blood indicate that for each of the first 2 weeks of chemotherapy,
the number of leukemic cells at the end of each week was 
1% of the
number at the beginning of the week, for the third week the number was
10%, and for the fourth and fifth weeks this number was
approximately unchanged.
Fig. 3
shows FS/W for marrow and blood when data were available for
both tissues in a patient for the same time interval. Where an actual
fractional survival could be calculated in both tissues, the data for
marrow and blood show excellent correlation (r = 0.91; n = 14; P < 0.001).
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and Fig. 2
provide some support for this latter possibility. In either
event, by day 21 the population of leukemic cells appeared to be
completely resistant to therapy with the four agents being used. If it
is assumed that at diagnosis 1011 to
1012 leukemic cells are present in the patient
with ALL, then the mean size of the resistant population can be
estimated as 106 to 107
cells, a substantial number. These observations are not consistent with the concept that the leukemic population is a homogeneous and unchanging collection of cells with a constant fraction being killed by each administration of chemotherapy, so that the rate of decrease is constant. They suggest that the biology of leukemia during treatment is more complex and that response to therapy changes with time.
Several factors may contribute to this complexity and changing response: (a) chemotherapy may directly mutate genes affecting drug resistance, but both the time delay between the occurrence of a mutation and its phenotypic expression, and the magnitude of the resistant population, make this explanation unlikely; (b) chemotherapy may lead to demethylation of the controlling regions of these genes and produce "epigenetic" mutations, but similar quantitative considerations make this possibility also unlikely; (c) quite possibly, chemotherapy may lead to increased expression of key genes in the leukemic cell by classical induction mechanisms. Such key genes might, for example, include genes controlling one or more of the following: proliferation, differentiation, apoptosis, drug transport, or drug metabolism; and (d) chemotherapy may lead to changes in expression of genes in nonleukemic cells, such as cells that are important in the detoxification or excretion of cytotoxic drugs.
Alternatively, the leukemic population evident after days 14–21 could already be present at diagnosis as one or more subpopulations that have various degrees of intrinsic drug resistance and that are selected out by therapy. Such subpopulation(s) could be resistant as the result of increased expression of nonmutated key genes but are more likely to have arisen as the result of genetic or epigenetic mutations that affect genes related to drug resistance and that result in subclonal evolution during growth of the leukemic clone. In their classic study, Luria and Delbruck (6) showed that apparent development of antibiotic resistance in bacterial populations was attributable to the selection of preexisting rare cells that had already mutated to antibiotic resistance prior to any antibiotic exposure and whose number was determined by the mutation rate and the point in time at which the mutation occurred. Goldie and Coldman (7) applied these concepts to cancer and considered the consequences of the presence of rare drug-resistant cancer cells. However, our data suggest that an additional factor may be present in leukemia and, by extension, in other cancers. Many of the mutations important in development or progression of leukemia are likely to affect genes involved in apoptosis. Such mutations will be selected for as they will lead to a survival advantage of and an increased mutation rate in the cells in which they occur, but they may incidentally result in drug resistance. As a consequence, a sizable population of drug-resistant cells is likely to develop within the leukemic population, before any exposure to drug. A previous study of three leukemic cell lines growing in vitro showed a mutation frequency of 3.1 x 10-5 to 1.1 x 10-3 during continuous culture and a mutation rate of 0.5–6.7 x 10-6 per cell division (8) . These values are quite consistent with the frequency of resistant cells that we observed in vivo.
These various inductive or selective mechanisms are not mutually
exclusive and, on the evidence available, it is difficult to decide
between them. There are, however, a number of circumstantial pieces of
evidence suggesting that genetically resistant cells may be present at
diagnosis and selected out by therapy: (a) the leukemic
population in ALL is often genetically heterogeneous at diagnosis as
evidenced by the presence of multiple immunoglobulin gene
rearrangements (9)
. Although gene rearrangements are
mutations that are selectively neutral, the coexistence of several
large subclones within the leukemic population is difficult to explain
unless one postulates that, except for the founder clone, there is
within each subclone a dominant sub-subclone that has resulted from a
mutation providing a selective growth advantage; (b) the
wide range in the size of the resistant population in different
patients (Table 1)
is well explained by Luria-Delbruck kinetics;
(c) the mutation rate of leukemic cells is consistent with
the observed frequency of drug-resistant cells; (d) at least
two patients have been observed that in each of whom there were two
genetically distinguishable leukemic populations present at diagnosis,
a major one that was drug sensitive and a minor one that was drug
resistant (10
, 11)
; (e) several patients have
been reported in whom there was, at diagnosis, a minor subpopulation of
leukemic cells that contained a mutation of p53, which presumptively
was associated with drug resistance (12
, 13)
.
Our observations on the kinetics of leukemia during induction therapy may explain several other observations on the biology of leukemia during treatment:
(a) Although most children with ALL are eventually cured, residual leukemia may be detected for quite long periods during postinduction treatment (14 , 15) . Although other explanations for this phenomenon are possible, one explanation is that a population of leukemic cells that shows resistance to the four drugs used during induction is likely to show cross-resistance to at least some of the drugs used subsequently and to therefore decline in number only slowly.
(b) Many workers have now observed that the size of the leukemic population at or shortly after the end of induction is a very strong predictor of cure or relapse (16, 17, 18, 19, 20, 21, 22, 23, 24) . The present study would interpret these observations as indicating that some cells resistant to the induction drugs are present in all or nearly all individuals and that it is the size of this resistant population, not its presence or its rate of decline, that determines outcome.
(c) Patients with a large percentage of blasts visible in the marrow on day 7 or day 14 are well recognized as having a poorer prognosis. This observation is often interpreted as indicating that such patients show a slower initial rate of decline of leukemia and that this slower rate of decline persists during subsequent treatment and leads to a poorer outcome. Our results suggest another interpretation, i.e., that if a large resistant population is present in a patient, then it may be the dominant population as early as day 14 or even day 7 and be responsible for a high blast percentage.
How might the size of the resistant population at the end of induction determine clinical outcome? Cure by chemotherapy depends on elimination of the leukemic clone before the development of completely drug-resistant cells. Complete drug resistance is likely to be multifactorial. Both inductive and selective mechanisms may play a role and each class of mechanism can explain the fact that the majority of patients are cured, whereas a minority relapse. However, irrespective of mechanism, it can be predicted that the size of the leukemic population at the end of induction will be an important determinant of the probability that completely drug-resistant cells are already present or will develop subsequently. Mutations are likely to play a necessary, even if not sufficient, role in complete drug resistance in most patients destined to relapse. The probability that mutations occur within a population is directly related to its size, its mutation rate, and the time for which it proliferates. Thus, a large "partially resistant" population is more likely than a small one to already contain or to develop the further mutations which confer complete resistance. Furthermore, the mutations of importance are more likely to occur before diagnosis, as the partially resistant population has been growing for some time, rather than during treatment, even though the mutation rate may have been increased by chemotherapy. We therefore suggest that the potential for cure with conventional therapy is usually already determined at diagnosis and depends on whether there are leukemic cells that already contain mutations that by themselves, or together with subsequent induced changes in gene expression, lead to complete drug resistance. The early presence of completely resistant leukemic cells may well explain the observation that high levels of MRD persisting in the early months of postinduction therapy indicate a very high probability of eventual relapse (23 , 24) .
Our findings also have implications for treatment of ALL. They suggest that the only effect of continuation or escalation of the dose of the same four induction drugs after day 14 is to diminish even further the size of the sensitive population. This effect is of lesser importance because the sensitive population has become the minor population. Depending on the mechanism(s) responsible for our findings, there may also be implications for the nature and/or scheduling of drug therapy.
|
FOOTNOTES |
|---|
1 This study was funded by the National Health and
Medical Research Council and Flinders 2000. M. J. B. received a
Rotary Peter Nelson Fellowship. 
2 To whom requests for reprints should be
addressed, at Hematology and Genetic Pathology, Flinders Medical
Center, Bedford Park, South Australia 5042, Australia.
3 The abbreviations used are: ALL, acute
lymphoblastic leukemia; MRD, minimal residual disease.
Received 11/ 3/00. Accepted 7/14/00.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
chain rearrangements. Leukemia (Baltimore), 7: 1066-1070, 1993.[Medline]
This article has been cited by other articles:
|
|
 |
|
  D. Bhojwani, H. Kang, R. X. Menezes, W. Yang, H. Sather, N. P. Moskowitz, D.-J. Min, J. W. Potter, R. Harvey, S. P. Hunger, et al. Gene Expression Signatures Predictive of Early Response and Outcome in High-Risk Childhood Acute Lymphoblastic Leukemia: A Children's Oncology Group Study J. Clin. Oncol., September 20, 2008; 26(27): 4376 - 4384. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Flotho, E. Coustan-Smith, D. Pei, C. Cheng, G. Song, C.-H. Pui, J. R. Downing, and D. Campana A set of genes that regulate cell proliferation predicts treatment outcome in childhood acute lymphoblastic leukemia Blood, August 15, 2007; 110(4): 1271 - 1277. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
, all samples) and
blood (•, all samples).
,
an upper limit was estimated for FS/W, because leukemia
was detected only in the first tissue sample;
,
a lower limit was estimated for FS/W because leukemia
was detected only in the second.