ABCB1 Variation Affects Myelosuppression,
Progression-free Survival and Overall Survival in
Paclitaxel/Carboplatin-treated Ovarian Cancer
Patients
Niclas Björn, Ingrid Jakobsen Falk, Ignace Vergote and Henrik Green
The self-archived postprint version of this journal article is available at Linköping
University Institutional Repository (DiVA):
http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-150852
N.B.: When citing this work, cite the original publication.
Björn, N., Jakobsen Falk, I., Vergote, I., Green, H., (2018), ABCB1 Variation Affects Myelosuppression, Progression-free Survival and Overall Survival in Paclitaxel/Carboplatin-treated Ovarian Cancer Patients, Basic & Clinical Pharmacology & Toxicology, 123(3), 277-287.
https://doi.org/10.1111/bcpt.12997
Original publication available at:
https://doi.org/10.1111/bcpt.12997
Copyright: Wiley (12 months)
http://eu.wiley.com/WileyCDA/
ABCB1 variation affects myelosuppression, progression-free 1
survival, and overall survival in paclitaxel/carboplatin-treated 2
ovarian cancer patients 3
Niclas Björna*, Ingrid Jakobsen Falka, Ignace Vergoteb, Henrik Gréena 4
a Clinical Pharmacology, Division of Drug Research, Department of 5
Medical and Health Sciences, Linköping University, SE-58185 6
Linköping, Sweden. 7
b Department of Obstetrics and Gynecology, University Hospital Leuven, 8
Leuven Cancer Institute, B-3000 Leuven, Belgium. 9
* Corresponding author: 10
Niclas Björn, Clinical Pharmacology, Division of Drug Research, 11
Department of Medical and Health Sciences, Linköping University, SE-12 58185 Linköping, Sweden. 13 E-mail: niclas.bjorn@liu.se 14 Telephone: +46101032029 15 Fax: +4613104195 16
Running title: ABCB1 in paclitaxel-treated ovarian cancer 17
Keywords: ABCB1, CYP2C8, adverse drug reactions, toxicity, 18
myelosuppression, ovarian cancer, paclitaxel. 19
Disclosure statements (conflicts of interest): The authors declare no 20
conflicts of interest. However, the study was conducted in collaboration 21
with Oasmia Pharmaceuticals AB, Uppsala, Sweden, who let us perform 22
this study on their phase III cohort and gave financial support to perform 23
the pharmacogenetics part of the study. 24
Abstract
25
The standard chemotherapy for ovarian cancer is paclitaxel/carboplatin. 26
Patients often exhibit myelosuppressive toxicity, and the treatment 27
response varies considerably. In this study, we investigated the 28
previously reported SNPs 1199G>A (rs2229109), 1236C>T (rs1128503), 29
2677G>T/A (rs2032582), 3435C>T (rs1045642) in ABCB1, and 30
1196A>G (rs10509681) in CYP2C8 and their association with treatment-31
induced myelosuppression, progression-free survival (PFS), and overall 32
survival (OS). From the Phase III study, OAS-07OVA, 525 patients (All) 33
treated with carboplatin and paclitaxel administered as Paclical (Arm A, 34
n=260) or Taxol® (Arm B, n=265), were included and genotyped using 35
pyrosequencing. Genotype associations with myelosuppression, PFS, and 36
OS were investigated using ANOVA, Kaplan-Meier analysis, and Cox 37
proportional hazard models. The most prominent finding was for the 38
ABCB1 variant 3435TT, which was significantly associated with 39
increased PFS in All (hazard ratio (HR) = 0.623), in Arm A (HR=0.590), 40
and in Arm B (HR=0.627), as well as increased OS in All (HR=0.443) 41
and in Arm A (HR=0.372) compared to the wildtype, 3435CC. For 42
toxicity, the most interesting finding concerned the haplotype, including 43
1236TT, 2677TT, and 3435TT, which was associated with higher 44
neutrophil values in Arm B (P=0.039) and less neutrophil decrease in All 45
(P=0.048) and in Arm B (P=0.021). It is noteworthy that the results 46
varied depending on the treatment arm which indicates that the effects of 47
ABCB1 variants vary with the treatment regimen. Our results reflect the 48
contradictory results of previous studies, confirming that small variations 49
in the composition of treatment regimens and patient populations may 50
influence the interpretation of SNPs effects on treatment outcome. 51
Introduction
52
Ovarian cancer is one of the leading causes of cancer-related deaths 53
among women in Europe [1]. The standard chemotherapy for primary 54
and recurrent ovarian cancer is paclitaxel in combination with 55
carboplatin [2-4]. Patients initially respond well to this treatment, but the 56
recurrence rate is around 70%, and the relative 5-year survival rate for 57
stages III and IV is only 25-30% [5]. Myelosuppression is a common 58
toxicity of paclitaxel and carboplatin therapy: neutrophil depression is 59
the most acute hematological toxicity, but thrombocytopenia, leukopenia, 60
and neutropenia are all commonly acquired by patients undergoing the 61
treatment [2, 6-9]. Neuropathy is another, often dose-limiting, side-effect 62
of paclitaxel [2, 10, 11]. 63
Paclitaxel is insoluble in water, and the solvent Cremophor® EL (CrEL) 64
is used in the traditional formulation of paclitaxel, Taxol®, to circumvent 65
this [12]. However, the use of CrEL can lead to hypersensitivity 66
reactions. Therefore, newer paclitaxel formulations that apply 67
nanotechnology to form water-soluble micelles, such as Paclical and 68
Abraxane® [12, 13], have been developed. 69
Currently, dosages are adapted to the body surface area and Calvert’s 70
formula for paclitaxel and carboplatin, respectively. However, there 71
remains considerable variation in both treatment response and toxicities. 72
There is a need for better means of individualization of the therapy. The 73
inter-individual differences in treatment response and toxicity are 74
thought to be partly attributable to genetic differences. 75
The two proteins, ABCB1 and CYP2C8, are involved in the 76
transportation and hepatic metabolism of paclitaxel, respectively [14]. 77
ABCB1 expression affects paclitaxel resistance in ovarian cancer cell 78
lines [15] and has for a long time been known to affect paclitaxel drug 79
disposition [16]. CYP2C8 is the main metabolizing enzyme of paclitaxel, 80
and genetic variations can alter this metabolism [17]. The effect of 81
variations in these genes on treatment response and toxicity in cancers 82
treated with paclitaxel has been extensively studied. However, the 83
results, from both our group and others have yielded inconclusive and 84
contradictory results [6, 7, 18-28], reviewed by Krens et al. [29] and 85
Fredericks et al. [30]. 86
This inconsistency and the inability to replicate previously reported 87
results for genetic variants in ABCB1 and CYP2C8 in relation to toxicity 88
and response can partly be blamed on the use of small sample sizes, 89
differences in treatment regimens, concomitant medications, the ethnicity 90
of patients, and the inclusion of various cancer types. This study was 91
designed to evaluate and validate the association between the SNPs 92
ABCB1 1199G>A (rs2229109), 1236C>T (rs1128503), 2677G>T/A 93
(rs2032582), 3435C>T (rs1045642), CYP2C8 1196A>G (rs10509681) 94
and paclitaxel/carboplatin-induced myelosuppression, progression-free 95
survival (PFS), and overall survival (OS) in 525 ovarian cancer patients 96
homogeneously treated with paclitaxel and carboplatin. 97
Materials and Methods
98
Study population
99
This study was conducted using a subset of patients from the phase III 100
study OAS-07OVA (EudraCT - 2008-002668-32, ClinicalTrials.gov - 101
NCT00989131). This phase III study included 789 patients with 102
recurrent ovarian cancer with the intent to compare the efficacy and 103
safety of paclitaxel micellar nanoparticles, Paclical, with paclitaxel 104
CrEL, Taxol®. Eligible patients were randomized to treatment Arm A 105
consisting of Paclical (250 mg/m2, given as a one-hour IV infusion) and 106
carboplatin (AUC 5-6) repeated every three weeks for six cycles; or 107
treatment Arm B consisting of Taxol® (175 mg/m2, given as a three-hour 108
IV infusion) and carboplatin (AUC 5-6) repeated every three weeks for 109
six cycles. A total of 525 patients (All; Arm A n=260 and Arm B n=265) 110
were included in this pharmacogenetics study after giving written 111
informed consent, following the Helsinki Declaration and with 112
permission from the regional ethics committees. Peripheral blood 113
samples were collected and stored at -80°C prior to DNA extraction. 114
Myelosuppression
115
Blood status was measured at baseline; at day 1, 8 and 15 through cycles 116
1 and 2; at day 1 and 8 through cycles 3-6; and at the end of the study. 117
Myelosuppression was evaluated for thrombocytes, leukocytes, and 118
neutrophils as nadir values during the first cycle and during the whole 119
treatment. The nadir values were graded according to the National 120
Cancer Institute (NCI) Common Terminology Criteria for Adverse 121
Events (CTCAE) version 4.03. The relative decreases in values from the 122
baseline were also evaluated during the first cycle and during the whole 123
treatment, according to Equation 1. 124
1:
125
Progression-free survival and overall survival
126
Progression-free survival (PFS) and overall survival (OS) were measured 127
as the time from inclusion to an event, i.e., confirmed progression with 128
Computed Tomography (CT) or death for PFS and OS, respectively. CT 129
scans were performed within 6 weeks before treatment, after cycle 3 and 130
6, every third month during the follow-up period, and when the patient 131
left the study. 132
DNA extraction and genotyping
133
DNA was extracted from blood samples using the Promega Maxwell 16 134
system (Promega Biotech, Nacka, Sweden) according to the 135
manufacturer’s protocol. Genotyping of 1199G>A, 1236C>T, 136
2677G>T/A, and 3435C>T in ABCB1 and 1196A>G in CYP2C8 was 137
performed using the pyrosequencer PSQ96MD (Qiagen, Uppsala, 138
Sweden) according to the manufacturer’s protocol, as previously 139
described [20, 22]. In short, HotStar Taq Master Mixture (VWR 140
International, Stockholm, Sweden) was used for polymerase chain 141
reaction (PCR) amplification performed on a Mastercycler gradient 142
(Eppendorf, Hamburg, Germany) with the total volume 10 µL using 0.4 143
µmol/L primer, 1.5 mmol/L MgCl2 and annealing temperature of 58°C. 144
Sequencing primer and biotinylated single-stranded PCR template were 145
annealed at 80°C for 2 minutes, enzyme and substrate were added, and 146
sequencing was carried out by dispensing dNTPs in a predefined order. 147
ABCB1 haplotype 148
The haplotype structure of ABCB1 has previously been described by 149
Kroetz et al. [31]. In this study, haplotypes involving the variants 150
1236C>T, 2677G>T (excluding the low frequency A variant) and 151
3435C>T in ABCB1 were investigated. Haplotypes were formed for all 152
patients that were homozygous wild-type, heterozygous or homozygous 153
variant at all three variant positions, denoted as CCGGCC, CTGTCT, 154
and TTTTTT (in order of physical appearance, i.e., 1236C>T, 2677G>T, 155
3435C>T) in the remainder of this paper. 156
Statistical analysis
157
Patient characteristics, baseline blood values, and genotype frequencies 158
were compared between the treatment arms using logistic regression or 159
ANOVA. Accordance of genotype distributions with the Hardy-160
Weinberg equilibrium and the Caucasian population in dbSNP was also 161
evaluated. 162
CTCAE grades of thrombocytes, leukocytes, and neutrophils were 163
compared between the treatment arms using Fisher’s exact test. 164
The association of baseline thrombocyte, leucocyte, and neutrophil 165
values with the patient characteristics treatment arm (A or B), age, 166
relapse (first or second), Eastern Cooperative Oncology Group (ECOG) 167
performance status, radiotherapy, cytoreductive surgery, previous 168
treatment with paclitaxel/docetaxel and chemotherapy-free interval (≥1 169
or <1 year, defined as the time from last chemotherapy course to 170
inclusion in this study) all relevant at the time of inclusion (tumor type 171
and tumor stage were evaluated at primary diagnosis and had likely 172
changed since; therefore, they were not investigated) were evaluated 173
using ANOVA. The genotype association with nadir values and relative 174
decreases of thrombocytes, leucocytes, and neutrophils (during both the 175
first cycle and the whole treatment) were assessed using ANOVA with 176
the respective baseline values (only for nadir values, not for relative 177
decreases) and treatment arm as covariables. 178
Kaplan-Meier analysis with the log-rank test was used to estimate 179
differences in PFS and OS between genotypes. For genotypes found to 180
be significant in the Kaplan-Meier analysis, Cox proportional hazard 181
models were constructed using the covariables treatment arm, age, 182
relapse, ECOG performance status, radiotherapy, cytoreductive surgery, 183
previous treatment with paclitaxel/docetaxel, chemotherapy-free interval 184
and dose delay/reduction (reduced or delayed dose administration during 185
treatment). Patients with no registered event (progression or death) were 186
censored at the date of their last CT scan or last follow-up for PFS and 187
OS, respectively. 188
IBM SPSS Statistics version (v.) 23 was used for logistic regressions and 189
Cox proportional hazard models. The statistical software R [32] (v. 3.2.0) 190
was used for Fisher’s exact test package stats v. 3.2.0), ANOVA (R-191
package stats v. 3.2.0), and Kaplan-Meier analysis (R-package survival 192
v. 2.38-3). All analyses were performed for All, Arm A, and Arm B. 193
Statistical significance was set as P≤0.05. 194
Results
195
Patient characteristics and baseline blood status
196
Patient characteristics are presented in Table 1. Logistic regression 197
showed no significant differences (P>0.05 for all comparisons) in patient 198
characteristics between the treatment arms, and there were no significant 199
differences between the treatment arms in terms of baseline blood values. 200
Furthermore, ANOVA revealed an association between patient 201
characteristics and baseline blood status (Supplemental Table 1). 202
However, these associations varied depending on blood cell types. All 203
patients were of Caucasian ethnicity, except one who was of African 204
descent. Although all the included patients were previously treated with 205
platinum-containing chemotherapy, at least six months had passed, so 206
they were deemed to be platinum sensitive. A clinical paper comparing 207
the treatment arms using the full study population of the phase III study 208
OAS-07OVA is under preparation. 209
Genotyping and genotype distribution
210
Genotyping was successful, and genotype frequencies are summarized in 211
Supplemental Table 2. There was no difference in genotype distributions 212
between the treatment arms (P>0.05 for all comparisons using logistic 213
regression). The distribution of genotypes corresponded to the 214
distribution seen for the Caucasian population in dbSNP (data not 215
shown) and was in accordance with the Hardy-Weinberg distribution 216
(data not shown). 217
The haplotype analysis of ABCB1 categorized 73, 169 and 89 patients as 218
CCGGCC, CTGTCT, and TTTTTT, respectively. This analysis excluded 219
294 patients who could not be classified into these haplotypes. 220
Toxicity
221
CTCAE grades for nadir values (first cycle and whole treatment) for 222
thrombocytes, leukocytes, and neutrophils are shown in Supplemental 223
Table 3. Fisher’s exact test showed significantly higher/greater toxicity 224
in treatment Arm A for all phenotypes (P<0.05 for all comparisons). 225
Inherent to the more prevalent toxicity in Arm A, 49 patients received 226
dose reductions/delay before the second treatment cycle whereas only 23 227
patients in Arm B required this. Furthermore, during the whole treatment 228
170 patients in Arm A and 137 in Arm B required dose 229
delays/reductions. 230
Results from the ANOVAs of 3435C>T, the haplotype and 231
myelosuppression are summarized in Table 2. In summary, the variant 232
alleles of 3435C>T and the haplotype were mainly associated with lower 233
toxicity (except in Arm A where the variant allele of 3435C>T was 234
associated with increased neutrophil toxicity). The direction of the 235
toxicity can be interpreted from boxplots in Supplemental Figure 1, 2 and 236
3 where mean values and standard deviations (SD) are also displayed for 237
all significant associations. The genotype 3435C>T was significant for 238
neutrophil whole treatment nadir in Arm A (P=0.018; mean count/L (SD) 239
for CC=8.5×108 (9.4×108), CT=6.0×108 (4.1×108) and TT=5.7×108 240
(4.7×108)), and the neutrophil first cycle relative decrease in All 241
(P=0.048; mean % (SD) for CC=70.0 (18.3), CT=61.8 (25.6) and 242
TT=63.2 (27.0)) and in Arm B (P=0.049; mean relative decrease (SD) for 243
CC=70.5 (17.5), CT=58.8 (26.6) and TT=60.1 (24.9)). The haplotype 244
was associated with the thrombocyte first cycle nadir in All (P=0.041; 245
mean count/L (SD) for CCGGCC=1.5×1011 (6.1×1010), 246
CTGTCT=1.8×1011 (7.6×1010) and TTTTTT=1.7×1011 (6.4×1010)), the 247
neutrophil first cycle nadir in Arm B (P=0.039; mean count/L (SD) for 248
CCGGCC=1.2×109 (8.6×108), CTGTCT= 1.6×109 (9.4×108) and 249
TTTTTT=1.5×109 (8.9×108)), the leukocyte whole treatment nadir in 250
Arm B (P=0.029; mean count/L (SD) for CCGGCC=2.1×109 (7.4×108), 251
CTGTCT=2.3×109 (8.2×108) and TTTTTT=2.6×109 (8.9×108)), and the 252
neutrophil first cycle relative decrease in All (P=0.048 mean % (SD) for 253
CCGGCC=70.0 (18.7), CTGTCT=61.7 (25.2) and TTTTTT=65.8 (19.9)) 254
and in Arm B (P=0.021; mean % (SD) for CCGGCC=71.4 (17.4), 255
CTGTCT=58.2 (25.2) and TTTTTT=61.1 (21.8)). It should be noted that 256
the covariable Treatment Arm (A or B) was always significant (data not 257
shown) in the ANOVAs of toxicity in All. This was expected due to the 258
harsher treatment regime in Arm A. Also, the covariable baseline blood 259
status was significant (data not shown) in all but one analysis (Arm A, 260
leukocytes, P=0.104). For toxicity associations with 1199G>A, 261
1236C>T, 2677G>T/A, and 1196A>G, see Supplemental Results and 262
Supplemental Table 4. 263
Progression-free survival
264
Kaplan-Meier analysis showed no initial differences in PFS between the 265
two treatment arms (data not shown). Table 3 lists the results from the 266
Meier analyses of PFS for 3435C>T and the haplotype. Kaplan-267
Meier curves are shown in Supplemental Figure 4 and mean PFS times in 268
months for all genotypes are listed in Supplemental Table 5. Cox 269
proportional hazard models for PFS are presented in Table 4. This 270
showed that the mean PFS in All for patients with 3435TT was 10.8 271
months compared to 9.6 months in patients with 3435CC, hazard ratio 272
(HR) = 0.623 (confidence interval (CI) = 0.464–0.835, P=0.002). In Arm 273
A, the mean PFS for patients with 3435TT was 11.6 months compared to 274
9.8 months in patients with 3435CC, HR=0.590 (CI=0.390–0.893, 275
P=0.013). In Arm B, the mean PFS for patients with 3435TT was 10.0 276
months compared to 9.3 months in patients with 3435CC, HR=0.627 277
(CI=0.409–0.960, P=0.032). For patients in All with the haplotype 278
TTTTTT, the mean PFS was 11.0 months compared to 9.9 months in 279
patients with the haplotype CCGGCC, HR=0.652 (CI=0.452–0.940, 280
P=0.022). This was also indicated in Arm A (P=0.057). For patients in 281
Arm B with the heterozygous haplotype CTGTCT, the mean PFS was 282
10.6 months compared to 10.1 months in patients with the haplotype 283
CCGGCC, HR=0.625 (CI=0.394–0.991, P=0.046). Kaplan-Meier 284
analyses of PFS for 1199G>A, 1236C>T, 2677G>T/A, and 1196A>G 285
are summarized in the Supplemental Results, Supplemental Table 5 and 286
6, and Supplemental Figure 5 (only 1236C>T). 287
Overall survival
288
Kaplan-Meier analysis showed no initial differences in OS between the 289
two treatment arms (data not shown). Results from the Kaplan-Meier 290
analyses of OS for 3435C>T and the haplotype are listed in Table 3 and 291
shown in Supplemental Figure 6, and the mean OS times in days for all 292
genotypes are presented in Supplemental Table 5. Cox proportional 293
hazard models for OS are presented in Table 5. This showed that the 294
mean OS in All for patients with 3435TT was 431 days compared to 374 295
days in patients with 3435CC, HR=0.443 (CI=0.264–0.746, P=0.002). In 296
Arm A, the mean OS for patients with 3435TT was 436 days compared 297
to 379 days in patients with 3435CC, HR=0.372 (CI=0.186–0.745, 298
P=0.005). This was not seen in Arm B (P=0.396). Similarly, the mean 299
OS in All for patients with the haplotype TTTTTT was 435 days 300
compared to 387 days in patients with the haplotype CCGGCC, 301
HR=0.467 (CI=0.248–0.878, P=0.018). In Arm A, the mean OS for the 302
haplotype TTTTTT was 444 days compared to 366 days for the 303
haplotype CCGGCC, HR=0.336 (CI=0.136–0.827, P=0.018). However, 304
this was not seen in Arm B (P=0.712). Kaplan-Meier analyses of OS for 305
1199G>A, 1236C>T, 2677G>T/A, and 1196A>G are summarized in the 306
Supplemental Results, Supplemental Table 5 and 6, and Supplemental 307
Figure 7 (for 1236C>T and 2677G>T). 308
Depiction of significant associations
309
Figure 1 shows an overall view of the significant associations for the 310
ABCB1 variants 1199G>A, 3435C>T and the haplotype, as well as the 311
CYP2C8 variant 1196A>G. 312
Discussion
313
The major strengths of this study are its large and clearly defined 314
population using homogenous treatment in two different regimens. A 315
potential drawback of the study is that variations in genes involved in 316
DNA repair, metabolism, and transportation essential for carboplatin [25, 317
33, 34] were not investigated. Variation in those genes could also 318
influence toxicity, PFS, and OS. However, as patients in both treatment 319
arms were treated with equal doses of carboplatin, the effects should be 320
similar in both and not a major contributor to the differences seen, 321
although their impact cannot be ruled out entirely. Also, no correction for 322
multiple testing was done, which should be borne in mind when 323
interpreting the results. 324
Toxicity
325
For ABCB1, the variant alleles of 1236C>T, 2677G>T, 3435C>T, and 326
the variant haplotype TTTTTT were associated with less 327
myelosuppressive toxicity in All and in Arm B. Although this was not 328
consistent for all blood cell types or for the duration of the treatment, it 329
indicates that the effects of ABCB1 SNPs on myelosuppression may 330
differ depending on factors specific to different blood cell types. 331
However, we did not see the same pattern in Arm A. At the same time, 332
Arm A had more myelosuppressive toxicity overall, probably due to the 333
higher concentration of paclitaxel. This indicates that the protective 334
effect of the variant alleles shown in Arm B can be counteracted by a 335
harsher treatment regimen like the one used in Arm A. 336
Hoffmeyer et al. [35] have previously shown decreased mRNA 337
expression for the variant allele of 3435C>T, followed by increased 338
digoxin plasma concentrations. This is in accordance with the results of 339
Wang et al. [36] which showed decreased expression and lowered 340
mRNA stability, which were subsequently supported by the findings of 341
Sissung et al. [26] describing increased neutropenia in patients carrying 342
3435C>T variant alleles. On the contrary, Baldissera et al. [37] describe 343
higher ABCB1 expression for 3435T allele carriers in hepatocellular 344
carcinoma samples. Furthermore, cell populations in peripheral blood 345
and bone marrow express ABCB1 [38]. The idea is that lowered 346
expression of ABCB1 leads to decreased cellular clearance of paclitaxel, 347
higher intracellular concentrations and thus more cell death and greater 348
toxicity. This infers that a low expression of ABCB1 in tumor cells would 349
be beneficial to the treatment of cancer, and high expression of ABCB1 350
in blood cells would be helpful for preventing toxicity. Our findings that 351
ABCB1 variant alleles decrease myelosuppressive toxicity cannot be 352
explained by the results of Hoffmeyer et al. [35], Wang et al. [36] and 353
Sissung et al. [26]. However, the expression of ABCB1 in different cell 354
types and the effects of the genetic variants on gene expression and 355
toxicity is not fully known. 356
The CYP2C8 variant 1196A>G was associated with higher leukocyte 357
toxicity in Arm A. 1196A>G is included in the CYP2C8*3 haplotype and 358
is in linkage with rs1113129 (D’=1.0 evaluated using SNAP [39]) in 359
CYP2C8-HapC. Previous studies have associated CYP2C8-HapC [40] 360
and CYP2C8*3 [18, 22] with decreased paclitaxel clearance and 361
CYP2C8-HapC [7] with lower leukocyte and neutrophil values. This 362
means that CYP2C8 variation could mediate an increased risk of 363
paclitaxel/carboplatin-induced leukopenia. 364
Survival
365
Chemotherapy-free interval ≥ 1 year was the biggest contributor to 366
longer PFS and OS. This is understandable since a more aggressive 367
cancer would result in a faster recurrence, and thus shorter 368
chemotherapy-free interval, and a worse prognosis. When PFS and OS 369
were investigated, 1236TT, 3435TT, and the haplotype TTTTTT were 370
associated with longer PFS; and 1236TT, 2677TT, 3435TT, and the 371
haplotype TTTTTT were associated with longer OS compared to 372
wildtypes. The most prominent finding was that 3435TT showed 373
increased PFS in All, Arm A, and Arm B as well as enhanced OS in All 374
and Arm A. This adds up to an indication of overall better prognosis for 375
patients being 3435TT. The meta-analysis of 4616 ovarian cancer 376
patients by Johnatty et al. [23] marginally associated 1236TT 377
(rs1128503) with improved OS and showed a tendency for a worse 378
prognosis in patients with higher tumor expression of ABCB1. Also, 379
ABCB1 2677T/A (allele carriers), and patients with 1236TT and 3435CC 380
have been associated with longer PFS [24, 27]. However, the latter is not 381
supported by our findings as we found the 3435TT variants to be 382
beneficial. Furthermore, some studies using different treatment regimens 383
have not seen any differences between genotypes and PFS or OS [19, 384
25], which might be explained by the fact that the different treatment 385
regimens in our study yielded different associations of genotypes for PFS 386
and OS (and myelosuppression). The effects observed are small and 387
likely not only dependent on the genotypes but also on treatment 388
conditions. 389
Another critical factor that could affect differences in survival is the 390
tumor expression of ABCB1. High ABCB1 tumor expression has been 391
associated with poor prognosis and chemotherapy resistance by Sun et 392
al. [41]. This was further strengthened by the results of Hoffmeyer et al. 393
[35] and Wang et al. [36] showing that 3435TT is associated with lower 394
ABCB1 expression, which supports our findings of 3435TT being 395
favorable for PFS and OS (unlike the association seen with toxicity, as 396
previously discussed). However, Gao et al. [42] have suggested that an 397
ABCB1-related survival difference in ovarian cancer patients is more 398
likely to be an effect of whole body paclitaxel clearance variations rather 399
than an effect of the variant on cancer cells. Since we do not know about 400
the somatic ABCB1 mutations or tumor expression of ABCB1 in our 401
samples, it is impossible to draw definite conclusions. 402
Lastly, the previously shown short progression-free survival of patients 403
with the ABCB1 variant 1199G>A [19, 21] was not seen in the presented 404
study. One explanation for this could be the presence of different 405
haplotypes in different study populations, yet another contributor to the 406
inconclusive body of literature published concerning ABCB1 genetic 407
variants and their importance for treatment effects. 408
Conclusion
409
Combining the results from myelosuppression, PFS and OS the study 410
indicates a toxicity protective effect of variant alleles in ABCB1, 411
especially for the variant allele of 3435C>T. The effects of ABCB1 412
variants in this study, using a large homogeneous patient sample, are 413
however small, and varied with the dose of paclitaxel. This currently 414
limits the predictive value of ABCB1 genotyping before the start of 415
paclitaxel treatment in clinical practice, based on our results. Our results 416
are reflected in the contradictory results of previous studies [6, 7, 18-30], 417
confirming that small variations in the composition of treatment 418
regimens, and the selection of patient populations could influence the 419
interpretation of the SNP effect on treatment outcomes. Furthermore, the 420
investigated variants might need different consideration depending on 421
whether the aim is to minimize toxicity or maximize the anti-tumor 422
effect of the treatment. Therefore, more thorough investigations of 423
paclitaxel treatments and genotypes are needed in combination with 424
ABCB1 expression analysis, and PK/PD studies in vivo and in vitro using 425
different cell types and both normal and tumor tissues to get a clear idea 426
of how to implement ABCB1 variation into the individualization of 427
treatment regimens containing paclitaxel. 428
Acknowledgements
429
The research presented in this article was supported by grants from the 430
Swedish Cancer Society, the Swedish Research Council, Linköping 431
University, and ALF grants Region Östergötland. Also, we would like to 432
thank Oasmia Pharmaceuticals AB, Uppsala, Sweden, for letting us 433
perform this pharmacogenetics study on their phase III cohort and for 434
their financial support of the pharmacogenetics part of the study. 435
References
436
1. Ferlay J, Parkin DM, Steliarova-Foucher E. Estimates of cancer 437
incidence and mortality in Europe in 2008. European journal of cancer. 438
2010;46:765-81. 439
2. du Bois A, Luck HJ, Meier W, Adams HP, Mobus V, Costa S, et 440
al. A randomized clinical trial of cisplatin/paclitaxel versus 441
carboplatin/paclitaxel as first-line treatment of ovarian cancer. J Natl 442
Cancer Inst. 2003;95:1320-9. 443
3. Parmar MK, Ledermann JA, Colombo N, du Bois A, Delaloye 444
JF, Kristensen GB, et al. Paclitaxel plus platinum-based chemotherapy 445
versus conventional platinum-based chemotherapy in women with 446
relapsed ovarian cancer: the ICON4/AGO-OVAR-2.2 trial. Lancet. 447
2003;361:2099-106. 448
4. Stuart GC, Kitchener H, Bacon M, duBois A, Friedlander M, 449
Ledermann J, et al. 2010 Gynecologic Cancer InterGroup (GCIG) 450
consensus statement on clinical trials in ovarian cancer: report from the 451
Fourth Ovarian Cancer Consensus Conference. Int J Gynecol Cancer. 452
2011;21:750-5. 453
5. Cannistra SA. Cancer of the ovary. The New England journal of 454
medicine. 2004;351:2519-29. 455
6. Bergmann TK, Brasch-Andersen C, Green H, Mirza MR, 456
Skougaard K, Wihl J, et al. Impact of ABCB1 variants on neutrophil 457
depression: a pharmacogenomic study of paclitaxel in 92 women with 458
ovarian cancer. Basic & clinical pharmacology & toxicology. 459
2012;110:199-204. 460
7. Green H, Khan MS, Jakobsen-Falk I, Avall-Lundqvist E, 461
Peterson C. Impact of CYP3A5*3 and CYP2C8-HapC on 462
paclitaxel/carboplatin-induced myelosuppression in patients with ovarian 463
cancer. Journal of pharmaceutical sciences. 2011;100:4205-9. 464
8. Langer CJ, Leighton JC, Comis RL, O'Dwyer PJ, McAleer CA, 465
Bonjo CA, et al. Paclitaxel and carboplatin in combination in the 466
treatment of advanced non-small-cell lung cancer: a phase II toxicity, 467
response, and survival analysis. J Clin Oncol. 1995;13:1860-70. 468
9. Baird RD, Tan DS, Kaye SB. Weekly paclitaxel in the treatment 469
of recurrent ovarian cancer. Nat Rev Clin Oncol. 2010;7:575-82. 470
10. Argyriou AA, Polychronopoulos P, Koutras A, Xiros N, Petsas 471
T, Argyriou K, et al. Clinical and electrophysiological features of 472
peripheral neuropathy induced by administration of cisplatin plus 473
paclitaxel-based chemotherapy. Eur J Cancer Care (Engl). 2007;16:231-474
7. 475
11. Kudlowitz D, Muggia F. Defining risks of taxane neuropathy: 476
insights from randomized clinical trials. Clin Cancer Res. 2013;19:4570-477
7. 478
12. Ma P, Mumper RJ. Paclitaxel Nano-Delivery Systems: A 479
Comprehensive Review. J Nanomed Nanotechnol. 2013;4:1000164. 480
13. Joerger M. Treatment regimens of classical and newer taxanes. 481
Cancer Chemother Pharmacol. 2016;77:221-33. 482
14. Oshiro C, Marsh S, McLeod H, Carrillo MW, Klein T, Altman 483
R. Taxane pathway. Pharmacogenet Genomics. 2009;19:979-83. 484
15. Vaidyanathan A, Sawers L, Gannon AL, Chakravarty P, Scott 485
AL, Bray SE, et al. ABCB1 (MDR1) induction defines a common 486
resistance mechanism in paclitaxel- and olaparib-resistant ovarian cancer 487
cells. Br J Cancer. 2016;115:431-41. 488
16. Gottesman MM, Pastan I, Ambudkar SV. P-glycoprotein and 489
multidrug resistance. Curr Opin Genet Dev. 1996;6:610-7. 490
17. Dai D, Zeldin DC, Blaisdell JA, Chanas B, Coulter SJ, 491
Ghanayem BI, et al. Polymorphisms in human CYP2C8 decrease 492
metabolism of the anticancer drug paclitaxel and arachidonic acid. 493
Pharmacogenetics. 2001;11:597-607. 494
18. Bergmann TK, Brasch-Andersen C, Green H, Mirza M, Pedersen 495
RS, Nielsen F, et al. Impact of CYP2C8*3 on paclitaxel clearance: a 496
population pharmacokinetic and pharmacogenomic study in 93 patients 497
with ovarian cancer. Pharmacogenomics J. 2011;11:113-20. 498
19. Bergmann TK, Green H, Brasch-Andersen C, Mirza MR, 499
Herrstedt J, Holund B, et al. Retrospective study of the impact of 500
pharmacogenetic variants on paclitaxel toxicity and survival in patients 501
with ovarian cancer. Eur J Clin Pharmacol. 2011;67:693-700. 502
20. Green H, Soderkvist P, Rosenberg P, Horvath G, Peterson C. 503
mdr-1 single nucleotide polymorphisms in ovarian cancer tissue: 504
G2677T/A correlates with response to paclitaxel chemotherapy. Clin 505
Cancer Res. 2006;12:854-9. 506
21. Green H, Soderkvist P, Rosenberg P, Horvath G, Peterson C. 507
ABCB1 G1199A polymorphism and ovarian cancer response to 508
paclitaxel. Journal of pharmaceutical sciences. 2008;97:2045-8. 509
22. Green H, Soderkvist P, Rosenberg P, Mirghani RA, Rymark P, 510
Lundqvist EA, et al. Pharmacogenetic studies of Paclitaxel in the 511
treatment of ovarian cancer. Basic & clinical pharmacology & 512
toxicology. 2009;104:130-7. 513
23. Johnatty SE, Beesley J, Gao B, Chen X, Lu Y, Law MH, et al. 514
ABCB1 (MDR1) polymorphisms and ovarian cancer progression and 515
survival: a comprehensive analysis from the Ovarian Cancer Association 516
Consortium and The Cancer Genome Atlas. Gynecologic oncology. 517
2013;131:8-14. 518
24. Johnatty SE, Beesley J, Paul J, Fereday S, Spurdle AB, Webb 519
PM, et al. ABCB1 (MDR 1) polymorphisms and progression-free 520
survival among women with ovarian cancer following 521
paclitaxel/carboplatin chemotherapy. Clin Cancer Res. 2008;14:5594-522
601. 523
25. Marsh S, Paul J, King CR, Gifford G, McLeod HL, Brown R. 524
Pharmacogenetic assessment of toxicity and outcome after platinum plus 525
taxane chemotherapy in ovarian cancer: the Scottish Randomised Trial in 526
Ovarian Cancer. J Clin Oncol. 2007;25:4528-35. 527
26. Sissung TM, Mross K, Steinberg SM, Behringer D, Figg WD, 528
Sparreboom A, et al. Association of ABCB1 genotypes with paclitaxel-529
mediated peripheral neuropathy and neutropenia. European journal of 530
cancer. 2006;42:2893-6. 531
27. Johnatty SE, Beesley J, Paul J, Fereday S, Spurdle AB, Webb 532
PM. ABCB1 (MDR1) Polymorphisms and Progression-Free Survival 533
among Women with Ovarian Cancer following Paclitaxel/Carboplatin 534
Chemotherapy (vol 14, pg 5594, 2008). Clinical Cancer Research. 535
2012;18:319-20. 536
28. Lambrechts S, Lambrechts D, Despierre E, Van Nieuwenhuysen 537
E, Smeets D, Debruyne PR, et al. Genetic variability in drug transport, 538
metabolism or DNA repair affecting toxicity of chemotherapy in ovarian 539
cancer. BMC Pharmacol Toxicol. 2015;16:2. 540
29. Krens SD, McLeod HL, Hertz DL. Pharmacogenetics, enzyme 541
probes and therapeutic drug monitoring as potential tools for 542
individualizing taxane therapy. Pharmacogenomics. 2013;14:555-74. 543
30. Frederiks CN, Lam SW, Guchelaar HJ, Boven E. Genetic 544
polymorphisms and paclitaxel- or docetaxel-induced toxicities: A 545
systematic review. Cancer Treat Rev. 2015;41:935-50. 546
31. Kroetz DL, Pauli-Magnus C, Hodges LM, Huang CC, 547
Kawamoto M, Johns SJ, et al. Sequence diversity and haplotype structure 548
in the human ABCB1 (MDR1, multidrug resistance transporter) gene. 549
Pharmacogenetics. 2003;13:481-94. 550
32. RCoreTeam. R: A language and environment for statistical 551
computing. R Foundation for Statistical Computing, Vienna, Austria; 552
2015. 553
33. Perez-Ramirez C, Canadas-Garre M, Alnatsha A, Villar E, 554
Delgado JR, Faus-Dader MJ, et al. Pharmacogenetic predictors of 555
toxicity to platinum based chemotherapy in non-small cell lung cancer 556
patients. Pharmacol Res. 2016;111:877-84. 557
34. Wu G, Jiang B, Liu X, Shen Y, Yang S. Association of GSTs 558
gene polymorphisms with treatment outcome of advanced non-small cell 559
lung cancer patients with cisplatin-based chemotherapy. Int J Clin Exp 560
Pathol. 2015;8:13346-52. 561
35. Hoffmeyer S, Burk O, von Richter O, Arnold HP, Brockmoller J, 562
Johne A, et al. Functional polymorphisms of the human multidrug-563
resistance gene: multiple sequence variations and correlation of one 564
allele with P-glycoprotein expression and activity in vivo. Proceedings of 565
the National Academy of Sciences of the United States of America. 566
2000;97:3473-8. 567
36. Wang D, Johnson AD, Papp AC, Kroetz DL, Sadee W. 568
Multidrug resistance polypeptide 1 (MDR1, ABCB1) variant 3435C>T 569
affects mRNA stability. Pharmacogenet Genomics. 2005;15:693-704. 570
37. Baldissera VD, de Mattos AA, Coral GP, de Araujo FB, Marroni 571
CA, de Mello Brandao AB, et al. Evaluation of the C3435T 572
polymorphism in the MDR1 gene in patients with hepatocellular 573
carcinoma. Ann Hepatol. 2012;11:899-906. 574
38. Drach D, Zhao S, Drach J, Mahadevia R, Gattringer C, Huber H, 575
et al. Subpopulations of normal peripheral blood and bone marrow cells 576
express a functional multidrug resistant phenotype. Blood. 577
1992;80:2729-34. 578
39. Johnson AD, Handsaker RE, Pulit SL, Nizzari MM, O'Donnell 579
CJ, de Bakker PIW. SNAP: a web-based tool for identification and 580
annotation of proxy SNPs using HapMap. Bioinformatics. 2008;24:2938-581
9. 582
40. Rodriguez-Antona C, Niemi M, Backman JT, Kajosaari LI, 583
Neuvonen PJ, Robledo M, et al. Characterization of novel CYP2C8 584
haplotypes and their contribution to paclitaxel and repaglinide 585
metabolism. Pharmacogenomics J. 2008;8:268-77. 586
41. Sun S, Cai J, Yang Q, Zhu Y, Zhao S, Wang Z. Prognostic Value 587
and Implication for Chemotherapy Treatment of ABCB1 in Epithelial 588
Ovarian Cancer: A Meta-Analysis. PLoS One. 2016;11:e0166058. 589
42. Gao B, Russell A, Beesley J, Chen XQ, Healey S, Henderson M, 590
et al. Paclitaxel sensitivity in relation to ABCB1 expression, efflux and 591
single nucleotide polymorphisms in ovarian cancer. Scientific reports. 592
2014;4:4669. 593
Figure
595596
Figure 1. Overall view of the significant associations in the study. All significant associations of the ABCB1 variants 1199G>A, 3435C>T, and
597
the haplotype# as well as the CYP2C8 SNP 1196A>G with thrombocytes, leukocytes, neutrophils, progression-free survival, and overall survival
598
for All, Arm A, and Arm B. Green indicates that the variant was associated with an advantage (lower toxicity, longer PFS, or longer OS) and 599
orange indicates that the variant was associated with a disadvantage (higher toxicity, shorter PFS, or shorter OS). 600
* p < 0.05. 601
** p < 0.01. 602
# Haplotype combining all homozygous wildtype (CCGGCC), heterozygous (CTGTCT), and homozygous variant (TTTTTT) at all three positions
603
1236C>T, 2677G>T, and 3435C>T. 604
Tables
605
Table 1: Patient characteristics for All patients, treatment Arm A and treatment Arm B. 606
All Arm A Arm B
N 525 260 265
Median age (range), at inclusion 56 (26-81) 56 (26-81) 56 (37-81)
Tumor type, primary diagnosis
Epithelial ovarian cancer 504 96.0% 253 97.3% 251 94.7%
Fallopian tube cancer 12 2.3% 4 1.5% 8 3.0%
Primary peritoneal cancer 9 1.7% 3 1.2% 6 2.3%
Tumor stage, primary diagnosis
Stage I 48 9.1% 25 9.6% 23 8.7%
Stage II 52 9.9% 26 10.0% 26 9.8%
Stage III 341 65.0% 171 65.8% 170 64.2%
Stage IV 83 15.8% 38 14.6% 45 17.0%
Missing 1 0.2% 0 0.0% 1 0.4%
Number of relapses, at inclusion
First relapse 399 76.0% 192 73.8% 207 78.1%
Second relapse 126 24.0% 68 26.2% 58 21.9%
ECOG Performance Status, at inclusion
0 291 55.4% 145 55.8% 146 55.1%
1 218 41.5% 106 40.8% 112 42.3%
2 16 3.0% 9 3.5% 7 2.6%
Radiotherapy, before inclusion
Yes 29 5.5% 14 5.4% 15 5.7%
No 496 94.5% 246 94.6% 250 94.3%
Cytoreductive surgery, before inclusion
Yes 482 91.8% 240 92.3% 242 91.3%
No 43 8.2% 20 7.7% 23 8.7%
Paclitaxel and/or docetaxel, before inclusion
Yes 270 51.4% 133 51.2% 137 51.7%
No 255 48.6% 127 48.8% 128 48.3%
Chemotherapy-free interval, time from last administrated
chemotherapy to inclusion
≥1 year 279 53.1% 139 53.5% 140 52.8%
<1 year 246 46.9% 121 46.5% 125 47.2%
Patient status, end of study
Dead 135 25.7% 67 25.8% 68 25.7%
Alive 390 74.3% 193 74.2% 197 74.3%
Abbreviation: ECOG, Eastern Cooperative Oncology Group. 607
Note: Logistic regression to assess differences in baseline characteristics between treatment Arm A and 608
treatment Arm B found no significant differences (P>0.05). 609
Table 2: P-values from ANOVA showing 3435C>T and ABCB1 haplotype effects on nadir values and relative decrease in thrombocytes,
610
leukocytes, and neutrophils during both the first cycle and the whole course of the treatment for All, Arm A, and Arm B. 611
All Arm A Arm B
Response variable Thrombocytes Leukocytes Neutrophils Thrombocytes Leukocytes Neutrophils Thrombocytes Leukocytes Neutrophils
Cycle 1 nadir 3435C>T 0.248 0.499 0.490 0.927 0.910 0.698 0.141 0.280 0.106 Haplotype# 0.041 0.263 0.275 0.363 0.979 0.856 0.102 0.094 0.039 Total nadir 3435C>T 0.444 0.597 0.466 0.705 0.487 0.018 0.343 0.305 0.297 Haplotype# 0.471 0.314 0.912 0.814 0.333 0.151 0.309 0.029 0.096
Cycle 1 relative decrease
3435C>T 0.417 0.109 0.048 0.943 0.364 0.558 0.182 0.290 0.049
Haplotype# 0.224 0.100 0.048 0.655 0.593 0.580 0.276 0.134 0.021
Total relative decrease
3435C>T 0.479 0.359 0.934 0.754 0.810 0.646 0.240 0.425 0.354
Haplotype# 0.563 0.094 0.660 0.850 0.462 0.353 0.272 0.060 0.055
Note: Significant P-values, P ≤ 0.05, are marked in bold. To understand the direction of toxicity for the significant associations see Supplemental 612
Figures 1, 2, and 3 with boxplots and mean values. 613
# Haplotype combining all homozygous wildtype (CCGGCC), heterozygous (CTGTCT), and homozygous variant (TTTTTT) at all three positions
614
1236C>T, 2677G>T, and 3435C>T. 615
Table 3: P-values for the Kaplan-Meier analyses of both progression-free survival and overall 616
survival for 3435C>T and the ABCB1 haplotype for All, Arm A, and Arm B. An overall P-value of 617
the difference as well as P-values comparing the individual genotypes with each other are shown. 618
Progression-free survival Overall survival Genotype P-value comparing All Arm A Arm B All Arm A Arm B 3435C>T overall difference 0.018 0.031 0.066 0.011 0.017 0.215
CC TT 0.006 0.032 0.019 0.003 0.006 0.299
CC CT 0.215 0.939 0.057 0.199 0.069 0.919
CT TT 0.039 0.012 0.761 0.025 0.171 0.103 Haplotype# overall difference 0.099 0.084 0.341 0.091 0.077 0.584
CCGGCC TTTTTT 0.052 0.125 0.134 0.043 0.025 0.585 CCGGCC CTGTCT 0.599 0.766 0.206 0.659 0.486 0.841
CTGTCT TTTTTT 0.057 0.021 0.910 0.033 0.088 0.313
Note: Significant P-values, P ≤ 0.05, are marked in bold. Mean progression-free survival and mean overall 619
survival are presented in Supplemental Table 5. 620
# Haplotype combining all homozygous wildtype (CCGGCC), heterozygous (CTGTCT), and homozygous 621
variant (TTTTTT) at all three positions 1236C>T, 2677G>T, and 3435C>T. 622
Table 4: Cox proportional hazard models of progression-free survival for the genotypes 1199G>A, 1236C>T, 3435C>T, and haplotype in
623
All, Arm A, and Arm B. The analysis was split into separate models for each significant genotype from the previous Kaplan-Meier
624
analyses of progression-free survival.
625
All Arm A Arm B
P HR 95% CI P HR 95% CI P HR 95% CI
Model with 1199G>A
Treatment Arm A 1 - - - B 0.025 1.270 1.031 - 1.564 - - - - Age 0.016 0.986 0.975 - 0.997 0.009 0.978 0.962 - 0.994 0.524 0.995 0.979 - 1.011 ECOG 0 1 1 1 1 0.447 1.086 0.878 - 1.342 0.951 1.01 0.731 - 1.396 0.524 1.097 0.824 - 1.461 2 0.013 2.243 1.187 - 4.240 0.477 1.372 0.574 - 3.278 <0.001 5.851 2.283 - 14.991 Relapse First 1 1 1 Second 0.590 1.070 0.836 - 1.371 0.279 1.226 0.847 - 1.774 0.996 0.999 0.710 - 1.405 Radiotherapy No 1 1 1 Yes 0.942 0.982 0.603 - 1.599 0.353 1.441 0.666 - 3.117 0.600 0.843 0.445 - 1.597 Cytoreductive surgery No 1 1 1 Yes 0.874 0.969 0.661 - 1.422 0.516 0.825 0.461 - 1.476 0.894 0.966 0.580 - 1.610 Paclitaxel/docetaxel No 1 1 1 Yes 0.884 0.984 0.792 - 1.222 0.363 0.855 0.611 - 1.197 0.753 1.048 0.783 - 1.402
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.548 0.446 - 0.674 0.003 0.631 0.465 - 0.857 <0.001 0.452 0.335 - 0.610 Dose delay/reduction No 1 1 1 Yes 0.233 0.880 0.714 - 1.085 0.384 0.864 0.622 - 1.200 0.204 0.832 0.627 - 1.105 1199G>A G/G 1 1 1 G/A + A/A 0.169 1.348 0.880 - 2.064 0.553 1.176 0.688 - 2.008 0.02 2.389 1.146 - 4.979 Model with 1236C>T Treatment Arm A 1 - - - B 0.026 1.265 1.028 - 1.557 - - - - Age 0.034 0.988 0.976 - 0.999 0.018 0.980 0.964 - 0.996 0.504 0.994 0.979 - 1.011
ECOG 0 1 1 1 1 0.440 1.087 0.879 - 1.343 0.883 0.976 0.706 - 1.349 0.562 1.089 0.817 - 1.451 2 0.020 2.132 1.125 - 4.040 0.655 1.223 0.506 - 2.953 <0.001 6.061 2.352 - 15.615 Relapse First 1 1 1 Second 0.567 1.075 0.839 - 1.377 0.395 1.177 0.809 - 1.713 0.923 1.017 0.724 - 1.429 Radiotherapy No 1 1 1 Yes 0.766 1.078 0.659 - 1.763 0.358 1.437 0.664 - 3.112 0.760 0.903 0.470 - 1.736 Cytoreductive surgery No 1 1 1 Yes 0.786 0.948 0.644 - 1.395 0.352 0.755 0.418 - 1.363 0.712 0.907 0.541 - 1.521 Paclitaxel/docetaxel No 1 1 1 Yes 0.838 0.978 0.788 - 1.214 0.243 0.818 0.584 - 1.146 0.764 1.046 0.781 - 1.401
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.534 0.434 - 0.657 0.001 0.6 0.440 - 0.820 <0.001 0.423 0.312 - 0.573 Dose delay/reduction No 1 1 1 Yes 0.118 0.845 0.683 - 1.044 0.372 0.863 0.624 - 1.193 0.100 0.784 0.588 - 1.047 1236C>T C/C 1 1 1 C/T 0.069 0.806 0.638 - 1.017 0.784 0.954 0.684 - 1.332 0.011 0.645 0.460 - 0.904 T/T 0.006 0.660 0.491 - 0.889 0.012 0.563 0.360 - 0.882 0.079 0.690 0.457 - 1.043 Model with 3435C>T Treatment Arm A 1 - - - B 0.031 1.256 1.021 - 1.546 - - - - Age 0.020 0.987 0.975 - 0.998 0.012 0.979 0.964 - 0.996 0.621 0.996 0.980 - 1.012 ECOG 0 1 1 1 1 0.335 1.110 0.898 - 1.372 0.985 1.003 0.723 - 1.392 0.585 1.083 0.812 - 1.445 2 0.017 2.180 1.147 - 4.145 0.725 1.173 0.482 - 2.851 <0.001 5.710 2.229 - 14.627 Relapse First 1 1 1 Second 0.609 1.066 0.834 - 1.362 0.376 1.180 0.818 - 1.704 0.981 1.004 0.716 - 1.407 Radiotherapy No 1 1 1 Yes 0.783 1.071 0.656 - 1.749 0.432 1.364 0.629 - 2.958 0.743 0.897 0.467 - 1.721 Cytoreductive surgery No 1 1 1 Yes 0.822 0.957 0.651 - 1.406 0.412 0.781 0.433 - 1.408 0.934 0.978 0.586 - 1.634 Paclitaxel/docetaxel No 1 1 1
Yes 0.772 0.969 0.780 - 1.202 0.237 0.817 0.584 - 1.143 0.805 1.037 .775 - 1.390
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.541 0.440 - 0.664 0.001 0.606 0.447 - 0.822 <0.001 0.438 0.324 - 0.593 Dose delay/reduction No 1 1 1 Yes 0.206 0.873 0.708 - 1.077 0.398 0.870 0.630 - 1.202 0.150 0.810 0.608 - 1.079 3435C>T C/C 1 1 1 C/T 0.039 0.756 0.580 - 0.986 0.684 0.926 0.638 - 1.344 0.009 0.599 0.407 - 0.881 T/T 0.002 0.623 0.464 - 0.835 0.013 0.590 0.390 - 0.893 0.032 0.627 0.409 - 0.960
Model with Haplotype#
Treatment Arm A 1 - - - B 0.155 1.210 0.930 - 1.575 - - - - Age 0.145 0.990 0.976 - 1.004 0.028 0.978 0.959 - 0.998 0.853 1.002 0.983 - 1.021 ECOG 0 1 1 1 1 0.238 1.171 0.901 - 1.520 0.298 1.246 0.824 - 1.885 0.925 1.017 0.713 - 1.450 2 0.391 1.510 0.590 - 3.867 0.662 0.714 0.157 - 3.235 0.012 4.785 1.416 - 16.175 Relapse First 1 1 1 Second 0.809 1.037 0.775 - 1.388 0.843 0.953 0.590 - 1.540 0.545 1.127 0.766 - 1.658 Radiotherapy No 1 1 1 Yes 0.977 1.008 0.577 - 1.762 0.62 1.241 0.529 - 2.913 0.610 0.817 0.375 - 1.777 Cytoreductive surgery No 1 1 1 Yes 0.584 1.134 0.722 - 1.782 0.942 0.973 0.458 - 2.064 0.598 1.171 0.650 - 2.110 Paclitaxel/docetaxel No 1 1 1 Yes 0.665 0.943 0.724 - 1.229 0.473 0.85 0.546 - 1.324 0.924 0.983 0.690 - 1.400
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.585 0.455 - 0.754 0.006 0.562 0.374 - 0.845 <0.001 0.503 0.348 - 0.728 Dose delay/reduction No 1 1 1 Yes 0.790 0.965 0.745 - 1.251 0.917 0.978 0.648 - 1.477 0.385 0.853 0.597 - 1.220 Haplotype# CCGGCC 1 1 1 CTGTCT 0.343 0.856 0.621 - 1.180 0.525 1.166 0.726 - 1.874 0.046 0.625 0.394 - 0.991 TTTTTT 0.022 0.652 0.452 - 0.940 0.057 0.594 0.347 - 1.015 0.125 0.662 0.391 - 1.121 Abbreviation: ECOG, Eastern Cooperative Oncology Group, HR, Hazard Ratio, 95% CI, 95% Confidence Interval.
Note: Significant P-values, P ≤ 0.05, are marked in bold. 627
#Haplotype combining all homozygous wildtype (CCGGCC), heterozygous (CTGTCT), and homozygous variant (TTTTTT) at all three positions
628
1236C>T, 2677G>T, and 3435C>T.
Table 5: Cox proportional hazard models of overall survival for the genotypes 1236C>T, 2677G>T, 3435C>T, and haplotype in All, Arm
630
A, and Arm B. The analysis was split into separate models for each significant genotype from the previous Kaplan-Meier analyses of
631
overall survival.
632
All Arm A Arm B
P HR 95% CI P HR 95% CI P HR 95% CI Model with 1236C>T Treatment Arm A 1 - - - B 0.636 0.919 0.647 - 1.304 - - - - - - Age 0.952 0.999 0.981 - 1.018 0.723 1.005 0.978 - 1.033 0.847 0.997 0.969 - 1.026 ECOG 0 1 1 1 1 0.096 1.372 0.946 - 1.989 0.525 1.196 0.688 - 2.081 0.139 1.484 0.880 - 2.502 2 0.215 1.822 0.706 - 4.703 0.567 1.450 0.406 - 5.184 0.303 2.307 0.470 - 11.329 Relapse First 1 1 1 Second 0.581 1.136 0.723 - 1.786 0.209 1.528 0.788 - 2.961 0.759 0.900 0.460 - 1.762 Radiotherapy No 1 1 1 Yes 0.832 1.121 0.392 - 3.205 0.406 1.860 0.431 - 8.020 0.725 0.760 0.165 - 3.494 Cytoreductive surgery No 1 1 1 Yes 0.476 0.794 0.422 - 1.496 0.962 0.975 0.335 - 2.834 0.347 0.670 0.290 - 1.544 Paclitaxel/docetaxel No 1 1 1 Yes 0.815 1.048 0.708 - 1.552 0.929 1.026 0.579 - 1.819 0.905 1.035 0.585 - 1.833
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.378 0.263 - 0.545 0.001 0.425 0.254 - 0.711 <0.001 0.336 0.192 - 0.587 Dose delay/reduction No 1 1 1 Yes 0.198 0.789 0.550 - 1.132 0.286 0.753 0.446 - 1.269 0.493 0.821 0.468 - 1.441 1236C>T C/C 1 1 1 C/T 0.448 0.857 0.576 - 1.277 0.365 0.776 0.448 - 1.344 0.677 0.872 0.458 - 1.660 T/T 0.007 0.487 0.288 - 0.823 0.019 0.390 0.178 - 0.855 0.189 0.585 0.263 - 1.301 Model with 2677G>T Treatment Arm A 1 - - -
B 0.526 0.891 0.622 - 1.274 - - - - - - Age 0.942 1.001 0.982 - 1.020 0.694 1.006 0.978 - 1.034 0.762 1.005 0.976 - 1.034 ECOG 0 1 1 1 1 0.113 1.362 0.930 - 1.995 0.512 1.215 0.679 - 2.173 0.138 1.493 0.879 - 2.536 2 0.239 1.909 0.651 - 5.598 0.608 1.517 0.309 - 7.452 0.525 1.689 0.335 - 8.508 Relapse First 1 1 1 Second 0.553 1.148 0.728 - 1.811 0.274 1.448 0.746 - 2.813 0.719 0.882 0.446 - 1.745 Radiotherapy No 1 1 1 Yes 0.796 1.150 0.400 - 3.308 0.496 1.666 0.384 - 7.227 0.774 0.796 0.168 - 3.763 Cytoreductive surgery No 1 1 1 Yes 0.742 0.896 0.466 - 1.722 0.956 1.031 0.350 - 3.038 0.66 0.823 0.346 - 1.960 Paclitaxel/docetaxel No 1 1 1 Yes 0.859 0.964 0.642 - 1.448 0.931 0.974 0.532 - 1.782 0.905 0.964 0.533 - 1.745
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.363 0.248 - 0.532 0.003 0.432 0.250 - 0.747 <0.001 0.316 0.177 - 0.564 Dose delay/reduction No 1 1 1 Yes 0.533 0.888 0.610 - 1.292 0.435 0.804 0.465 - 1.390 0.991 1.003 0.561 - 1.796 2677G>T G/G 1 1 1 G/T 0.672 0.913 0.598 - 1.393 0.531 0.830 0.464 - 1.487 0.809 1.089 0.547 - 2.166 T/T 0.006 0.472 0.275 - 0.810 0.018 0.382 0.172 - 0.850 0.292 0.637 0.276 - 1.473 Model with 3435C>T Treatment Arm A 1 - - - B 0.629 0.917 0.645 - 1.303 - - - - - - Age 0.867 0.998 0.980 - 1.017 0.925 1.001 0.976 - 1.028 0.880 0.998 0.971 - 1.026 ECOG 0 1 1 1 1 0.073 1.411 0.968 - 2.056 0.282 1.363 0.775 - 2.395 0.182 1.428 0.846 - 2.411 2 0.273 1.699 0.658 - 4.386 0.599 1.402 0.398 - 4.944 0.346 2.121 0.444 - 10.132 Relapse First 1 1 1 Second 0.496 1.165 0.750 - 1.809 0.184 1.537 0.815 - 2.900 0.735 0.891 0.459 - 1.732 Radiotherapy No 1 1 1 Yes 0.816 1.133 0.396 - 3.238 0.367 1.964 0.454 - 8.498 0.751 0.780 0.168 - 3.617 Cytoreductive surgery No 1 1 1
Yes 0.626 0.855 0.454 - 1.608 0.965 1.023 0.364 - 2.875 0.429 0.714 0.310 - 1.646
Paclitaxel/docetaxel No 1 1 1
Yes 0.852 1.038 0.699 - 1.542 0.887 1.043 0.587 - 1.852 0.84 1.062 0.592 - 1.904
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.404 0.282 - 0.579 0.001 0.413 0.247 - 0.690 0.001 0.375 0.214 - 0.657 Dose delay/reduction No 1 1 1 Yes 0.219 0.797 0.556 - 1.144 0.326 0.769 0.455 - 1.300 0.623 0.866 0.489 - 1.534 3435C>T C/C 1 1 1 C/T 0.079 0.674 0.434 - 1.047 0.033 0.530 0.295 - 0.951 0.868 0.933 0.414 - 2.103 T/T 0.002 0.443 0.264 - 0.746 0.005 0.372 0.186 - 0.745 0.396 0.672 0.268 - 1.684
Model with Haplotype#
Treatment Arm A 1 - - - B 0.598 0.890 0.576 - 1.374 - - - - - - Age 0.834 1.003 0.979 - 1.026 0.776 1.005 0.972 - 1.039 0.861 1.003 0.969 - 1.038 ECOG 0 1 1 1 1 0.255 1.299 0.827 - 2.041 0.332 1.422 0.699 - 2.892 0.374 1.319 0.716 - 2.430 2 0.421 1.681 0.475 - 5.949 0.823 1.307 0.126 - 13.604 0.769 1.292 0.235 - 7.111 Relapse First 1 1 1 Second 0.623 0.878 0.524 - 1.474 0.867 0.931 0.406 - 2.138 0.45 0.751 0.357 - 1.579 Radiotherapy No 1 1 1 Yes 0.738 0.782 0.184 - 3.315 0.332 2.127 0.463 - 9.778 0.978 0.000 0.000 - inf Cytoreductive surgery No 1 1 1 Yes 0.835 1.083 0.511 - 2.294 0.936 0.948 0.257 - 3.494 0.808 1.136 0.405 - 3.187 Paclitaxel/docetaxel No 1 1 1 Yes 0.690 1.102 0.684 - 1.776 0.436 1.352 0.634 - 2.885 0.648 0.853 0.430 - 1.692
Chemotherapy-free interval <1 year 1 1 1
≥1 year <0.001 0.406 0.261 - 0.631 0.013 0.418 0.210 - 0.833 0.006 0.380 0.190 - 0.760 Dose delay/reduction No 1 1 1 Yes 0.414 0.830 0.531 - 1.297 0.667 0.857 0.424 - 1.730 0.619 0.838 0.418 - 1.682 Haplotype# CCGGCC 1 1 1 CTGTCT 0.629 0.874 0.507 - 1.508 0.356 0.703 0.332 - 1.487 0.916 1.050 0.424 - 2.602 TTTTTT 0.018 0.467 0.248 - 0.878 0.018 0.336 0.136 - 0.827 0.712 0.826 0.298 - 2.286
Abbreviation: ECOG, Eastern Cooperative Oncology Group, HR, Hazard Ratio, 95% CI, 95% Confidence Interval, inf, infinite. 633
Note: Significant P-values, P ≤ 0.05, are marked in bold. 634
# Haplotype combining all homozygous wildtype (CCGGCC), heterozygous (CTGTCT) and homozygous variant (TTTTTT) at all three positions
635
1236C>T, 2677G>T, and 3435C>T. 636
Supplemental results, figures, and tables
Title:
ABCB1 variation affects myelosuppression, progression-free survival, and overall survival in paclitaxel/carboplatin-treated ovarian cancer patients
Journal:
Basic & Clinical Pharmacology & Toxicology
Authors:
Niclas Björna*, Ingrid Jakobsen Falka, Ignace Vergoteb and Henrik Gréena
Affiliations:
a Clinical Pharmacology, Division of Drug Research, Department of Medical and Health Sciences, Linköping University, SE-58185 Linköping, Sweden
b Department of Obstetrics and Gynecology, University Hospital Leuven, Leuven Cancer Institute, B-3000 Leuven, Belgium
* Corresponding author:
Niclas Björn, Clinical Pharmacology, Division of Drug Research, Department of Medical and Health Sciences, Linköping University, SE-58185 Linköping, Sweden.
Supplemental results
Results for ABCB1 SNPs 1199G>A, 1236C>T, and 2677G>T/A and the CYP2C8 SNP 1196A>G are here presented. Supplemental Figures 1-7 and Supplemental Tables 1-6 will follow the summarizing text below.
Genotyping and genotype distribution
All SNPs were successfully genotyped in all patients and genotype frequencies are summarized in Supplemental Table 2. Due to the low number of A/A genotypes for 1199G>A these were analyzed together with heterozygous genotypes G/A in a dominant manner. Likewise, the G/G genotype of 1196A>G was analyzed together with the A/G genotypes. For 2677G>T/A genotypes including the A allele was excluded from further studies because of the low genotype frequency, leaving 491 patients for analyzes of 2677G>T.
Toxicity
The association of 1199G>A, 1236C>T, 2677G>T/A, and 1196A>G with nadir values and relative decrease of thrombocytes, leukocytes, and neutrophils during the first cycle and the whole treatment were evaluated with ANOVA and the results are listed in Supplemental Table 4. Supplemental Figure 1, 2, and 3 can be used to determine that the significant associations show lower toxicity for the variant alleles of 1236C>T and 2677G>T whereas the variant alleles of 1199G>A and 1196A>G show increased toxicity.
Progression free survival and overall survival
Supplemental Table 7 lists all P-values from the Kaplan-Meier analysis of PFS and OS for 1199G>A, 1236C>T, 2677G>T/A, and 1196A>G. Variants that had at least one significant P-value (1199G>A, 1236C>T, and 2677G>T) were further analyzed using Cox proportional hazard models, results are listed in Table 4 and 5 in the main article for PFS and OS, respectively. These analyses showed that variant alleles of 1236C>T had prolonged PFS in All, Arm A and Arm B. Kaplan-Meier curves for PFS and 1236C>T are shown in Supplemental Figure 5. Further, patients being 1236TT or 2677TT showed prolonged OS in All and Arm A compared to wildtype. OS Kaplan-Meier curves for 1236C>T and 2677G>T are shown in Supplemental Figure 7. Mean PFS and OS times for the variants are listed in Supplemental Table 5.
Supplemental Figure 1. The figures show jittering of the nadir values during the first cycle for different genotypes (with significant p-values in the ANOVA displayed in Table 2 and Supplementary Table 4)accompanied by boxplots, mean, and standard deviations (SD): A) Leukocyte count for Arm B shown for each of the 1199G>A genotypes G/A + A/A and G/G; B) Thrombocyte count for All shown for each of the 1236C>T genotypes C/C, C/T and T/T; C) Thrombocyte count for All divided on 2677G>T genotypes G/G, G/T and T/T; D) Thrombocyte count for All shown for each of the ABCB1 haplotypes# CCGGCC, CTGTCT, and TTTTTT; E) Neutrophil count for Arm B shown for each of the ABCB1 haplotypes# CCGGCC, CTGTCT, and TTTTTT.
Supplemental Figure 2. The figures show a jittering of the nadir values during the whole course of the treatment for different genotypes (with significant p-values in the ANOVA displayed in Table 2 and Supplementary Table 4) accompanied by boxplots, mean, and standard deviations (SD): A) Leukocyte count for Arm B shown for each of the 1199G>A genotypes G/A + A/A and G/G; B) Neutrophil count for Arm B shown for each of the 2677G>T
genotypes G/G, G/T and T/T; C) Neutrophil count for Arm A shown for each of the 3435C>T genotypes C/C, C/T and T/T; D) Leukocyte count for Arm B shown for each of the ABCB1 haplotypes# CCGGCC, CTGTCT and TTTTTT; E) Leukocyte count for Arm A shown for each of the CYP2C8 1196A>G genotypes A/A and A/G + G/G.
Supplemental Figure 3. The figures show a jittering of the relative decrease in nadir values to the baseline values for either the first cycle (A-D) or the whole course of the treatment (E and F) for different genotypes (with significant p-values in the ANOVA displayed in Table 2 and Supplementary Table 4) accompanied by boxplots, mean, and standard deviations (SD): A) Neutrophil relative decrease first cycle for All shown for each of the 3435C>T genotypes C/C, C/T and T/T; B) Neutrophil relative decrease first cycle for Arm B shown for each of the 3435C>T genotypes C/C, C/T and T/T; C) Neutrophil relative decrease first cycle for All shown for each of the ABCB1 haplotypes# CCGGCC, CTGTCT and TTTTTT; D) Neutrophil relative decrease first cycle for Arm B shown for each of the ABCB1 haplotypes# CCGGCC, CTGTCT and TTTTTT; E) Neutrophil relative decrease during the, whole treatment for Arm B shown for each of the 2677G>T genotypes G/G, G/T and T/T; F) Leukocyte relative decrease during the whole treatment for Arm A shown for each of the CYP2C8 1196A>G genotypes A/A
Supplemental Figure 4. Kaplan-Meier curves for progression-free survival (PFS), in months, for 3435C>T and the haplotype in All, Arm A and Arm B. Significance was assessed using the log-rank test. In All, patients with 3435TT had significantly longer PFS compared to 3435CC (P = 0.006) (A) this was also seen in Arm A (P = 0.032) (B) and Arm B (P = 0.019) (C). For the haplotype#, PFS was significantly longer for patients with TTTTTT compared to CTGTCT in Arm A (P = 0.021) (E), but not in all (P > 0.05) (D) and Arm B (P > 0.05) (F). Mean PFS times for the genotypes and groups are listed in Supplemental Table 5. # Haplotype combining homozygous wildtype (CCGGCC), heterozygous (CTGTCT), and homozygous variant (TTTTTT) at all three positions 1236C>T, 2677G>T, and 3435C>T.
