Articles
Mechanisms of peripheral neuropathy associated with
bortezomib and vincristine in patients with newly
diagnosed multiple myeloma: a prospective analysis of
data from the HOVON-65/GMMG-HD4 trial
Annemiek Broyl*, Sophie L Corthals*, Joost L M Jongen, Bronno van der Holt, Rowan Kuiper, Yvonne de Knegt, Mark van Duin, Laila el Jarari,
Uta Bertsch, Henk M Lokhorst, Brian G Durie, Hartmut Goldschmidt, Pieter Sonneveld
Summary
Background Bortezomib-induced peripheral neuropathy is a dose-limiting toxicity in patients with multiple myeloma, Lancet Oncol 2010; 11: 1057­65
often requiring adjustment of treatment and affecting quality of life. We investigated the molecular profiles of early- Published Online
September 22, 2010
onset (within one treatment cycle) versus late-onset (after two or three treatment cycles) bortezomib-induced
DOI:10.1016/S1470-
peripheral neuropathy and compared them with those of vincristine-induced peripheral neuropathy during the 2045(10)70206-0
induction phase of a prospective phase 3 trial.
See Reflection and Reaction
page 1014
Methods In the induction phase of the HOVON-65/GMMG-HD4 trial, patients (aged 18­65 years) with newly See Review page 1086
diagnosed Salmon and Durie stage 2 or 3 multiple myeloma were randomly assigned to three cycles of bortezomib- *Contributedequallytothis
based or vincristine-based induction treatment. We analysed the gene expression profiles and single-nucleotide paper
polymorphisms (SNPs) of pretreatment samples of myeloma plasma cells and peripheral blood, respectively. This Department of Haematology
study is registered, number ISRCTN64455289.
(A Broyl MD, S L Corthals MSc,
R Kuiper BSc, Y de Knegt BSc,
M van Duin PhD,
Findings We analysed gene expression profiles of myeloma plasma cells from 329 (39%) of 833 patients at diagnosis, ProfPSonneveldMD),
and SNPs in DNA samples from 369 (44%) patients. Early-onset bortezomib-induced peripheral neuropathy was Department of Neurology
noted in 20 (8%) patients, and 63 (25%) developed the late-onset type. Early-onset and late-onset vincristine-induced (J L M Jongen MD), and HOVON
Data Centre (B van der Holt PhD,
peripheral neuropathy was noted in 11 (4%) and 17 (7%) patients, respectively. Significant genes in myeloma plasma LelJarariMSc),ErasmusMC,
cells from patients that were associated with early-onset bortezomib-induced peripheral neuropathy were the enzyme Rotterdam,Netherlands;
coding genes RHOBTB2 (upregulated by 1·59 times; p=4·5×10­), involved in drug-induced apoptosis, CPT1C Department of Internal
(1·44 times; p=2·9×10­), involved in mitochondrial dysfunction, and SOX8 (1·68 times; p=4·28×10­¹³), involved in Medicine V, University of
Heidelberg, Heidelberg,
development of peripheral nervous system. Significant SNPs in the same patients included those located in the Germany(UBertschMD,
apoptosis gene caspase 9 (odds ratio [OR] 3·59, 95% CI 1·59­8·14; p=2·9×10­³), ALOX12 (3·50, 1·47­8·32; p=3·8×10­³), ProfHGoldschmidtMD);
and IGF1R (0·22, 0·07­0·77; p=8·3×10­³). In late-onset bortezomib-induced peripheral neuropathy, the significant Department of Haematology,
genes were SOD2 (upregulated by 1·18 times; p=9·6×10­³) and MYO5A (1·93 times; p=3·2×10­²), involved in University Medical Centre,
Utrecht, Netherlands
development and function of the nervous system. Significant SNPs were noted in inflammatory genes MBL2 (ProfHMLokhorstMD);and
(OR 0·49, 95% CI 0·26­0·94; p=3·0×10­²) and PPARD (0·35, 0·15­0·83; p=9·1×10­³), and DNA repair genes ERCC4 Cedars-SinaiComprehensive
(2·74, 1·56­4·84; p=1·0×10­³) and ERCC3 (1·26, 0·75­2·12; p=3·3×10­³). By contrast, early-onset vincristine-induced Cancer Center, Los Angeles, CA,
peripheral neuropathy was characterised by upregulation of genes involved in cell cycle and proliferation, including USA (B G Durie MD)
AURKA (3·31 times; p=1·04×10­²) and MKI67 (3·66 times; p=1·82×10­³), and the presence of SNPs in genes involved Correspondence to:
Dr Annemiek Broyl, Department
in these processes--eg, GLI1 (rs2228224 [0·13, 0·02­0·97, p=1·18×10­²] and rs2242578 [0·14, 0·02­1·12, p=3·00×10­²]). ofHaematology,ErasmusMC,
Late-onset vincristine-induced peripheral neuropathy was associated with the presence of SNPs in genes involved in Dr Molewaterplein 50,
absorption, distribution, metabolism, and excretion--eg, rs1413239 in DPYD (3·29, 1·47­7·37, 5·40×10­³) and 3015 GE Rotterdam,
rs3887412 in ABCC1 (3·36, 1·47­7·67, p=5·70×10­³).
Netherlands
a.broyl@erasmusmc.nl
Interpretation Our results strongly suggest an interaction between myeloma-related factors and the patient's genetic
background in the development of treatment-induced peripheral neuropathy, with different molecular pathways
being implicated in bortezomib-induced and vincristine-induced peripheral neuropathy.
Funding German Federal Ministry of Education and Research, Dutch Cancer Foundation Queen Wilhelmina, European
Hematology Association, International Myeloma Foundation, Erasmus MC, and Janssen-Cilag Orthobiotech.
Introduction
in combination with conventional drugs resulted in
Bortezomib (Millennium Pharmaceuticals, Cambridge,
high rates of complete response and very good partial
MA, USA) is a boronic acid dipeptide, which specifically
response.1­4 This drug is generally well tolerated;
inhibits the chymotryptic site of the 26S proteasome.
however, one of its most frequent and potentially
In patients with newly diagnosed myeloma, bortezomib
disabling side-effects is the development of a painful,
www.thelancet.com/oncology Vol 11 November 2010
1057
Articles
sensory peripheral neuropathy,5­7 often requiring dose
Methods
modification or discontinuation of bortezomib, which
Patients
negatively affects clinical endpoints and quality of life.8
833 patients (aged 18­65 years) with newly diagnosed
Grade
1
and
2
bortezomib-induced
peripheral
Salmon and Durie stage 2­3 multiple myeloma were
neuropathy can arise in 27­75% of patients with
enrolled in a prospective, randomised phase 3 trial
recurrent multiple myeloma and in 25­33% of those
(HOVON-65/GMMG-HD4; EudraCTnr2004-000944-26)
with newly diagnosed multiple myeloma, whereas
in 75 centres in the Netherlands, Germany, and Belgium.3
grade 3 and 4 peripheral neuropathy might affect 0­30%
Patients were excluded if they had amyloidosis or
of patients with recurrent disease and 0­18% of those
monoclonal gammopathy of unknown significance, and
with newly diagnosed disease.9 In most patients, this
baseline peripheral neuropathy of grade 2 or more.
side-effect is reversible and does not seem to be affected
The trial was done in accordance with the Declaration
by the number or type of previous treatments.7
of Helsinki, and was approved by a medical ethics
Bortezomib-induced peripheral neuropathy results
review committee. We obtained written informed
from axonal degeneration,10,11 often occurring within the
consent from the patients for treatment and sample
first cycles of treatment, and does not seem to increase
procurement.
after the fifth cycle of bortezomib.7
Little is known about the mechanism of bortezomib-
Procedures
induced peripheral neuropathy, but a multifactorial
Patients were randomly assigned to three cycles of
pathogenesis seems likely. Damage to mitochondria
induction treatment with vincristine 0·4 mg intra-
and endoplasmic reticulum through activation of
venously on days 1­4, doxorubicin 9 mg/m² intravenously
apoptosis has been seen in dorsal root ganglia of mice
on days 1­4, and dexamethasone 40 mg orally on days
given bortezomib.11 Additionally, mechanisms such as
1­4, 9­12, and 17­20 or bortezomib 1·3 mg/m²
dysregulation of mitochondrial calcium homoeostasis,12
intravenously on days 1, 4, 8, and 11, doxorubicin
autoimmune factors and inflammation,13 and blockade
9 mg/m² intravenously on days 1­4, and dexamethasone
of
nerve-growth-factor-mediated
neuronal
survival
40 mg orally on days 1­4, 9­12, and 17­20. Stem-cells
through inhibition of the activation of nuclear factor B
were mobilised by use of cyclophosphamide 1000 mg/m²
(NFB)6
could
contribute
to
bortezomib-induced
intravenously on day 1, doxorubicin 15 mg/m²
peripheral neuropathy. Evidence that multiple myeloma
intravenously on days 1­4, dexamethasone 40 mg orally
is also implicated in peripheral neuropathy was
on days 1­4, and granulocyte colony-stimulating factor
described by Ropper and Gorson14 in 1998. Baseline
(filgrastim) 10 g/kg per day subcutaneously, divided in
neuropathy is present in 15­20% of patients with newly
two doses per day, from day 5 until last stem cell
diagnosed myeloma,15,16 which might be of both axonal
collection. After induction therapy, patients received one
and demyelinating subtypes.14 The role of myeloma-
or two cycles of high-dose melphalan (200 mg/m²
related factors in peripheral neuropathy related to
intravenously) with autologous stem-cell rescue followed
treatment is not clear. Bortezomib-induced peripheral
by maintenance treatment with thalidomide (50 mg per
neuropathy was noted at higher frequencies in patients
day orally; group assigned to vincristine-based induction
with multiple myeloma than in those with solid
treatment) or bortezomib (1·3 mg/m² intravenously
tumours.17 Richardson and colleagues16 characterised
once every 2 weeks; group assigned to bortezomib-based
the possible role of myeloma-related factors in
induction treatment) for 2 years. Treatment was not
bortezomib-induced
peripheral
neuropathy
using
masked for physicians and patients.
plasma cells from patients with multiple myeloma.
Severity of neuropathy was graded at baseline and after
Additionally, we have noted that inherited single-
each treatment cycle by use of the National Cancer
nucleotide polymorphisms (SNPs) are associated with a
Institute's Common Toxicity Criteria for Adverse Events
higher probability of developing thalidomide-induced
criteria (version 3.0).18 All data were analysed centrally.
or
bortezomib-induced
peripheral
neuropathy
No neurological assessment was undertaken to objectify
(Corthals SL, unpublished data). We therefore analysed
peripheral
neuropathy.
Since
grade
1
peripheral
myeloma-related gene expression and inherited patient
neuropathy could easily be missed or misinterpreted,
variations as indicators of the potential risk of
and because it does not include pain or interfere with the
developing treatment-related peripheral neuropathy.
activities of daily life, we decided that grade 1 peripheral
We investigated whether particular molecular profiles
neuropathy was not clinically significant enough for the
were
specific
for
early-onset
versus
late-onset
molecular analysis and therefore cases of this grade were
bortezomib-induced
peripheral
neuropathy
and
excluded. Furthermore, the dose-modification guidelines
compared these with genetic profiles associated with
established during the SUMMIT,6 CREST,5 and APEX19
early-onset
versus
late-onset
vincristine-induced
trials did not recommend discontinuation of bortezomib
peripheral neuropathy to elucidate molecular differences
or dose modifications when grade 1 bortezomib-induced
associated with the development of peripheral neuro-
peripheral neuropathy occurred. We did not routinely
pathy after the different treatments.
assess data for diabetes and vascular disease. Development
1058
www.thelancet.com/oncology Vol 11 November 2010
Articles
of peripheral neuropathy after the first cycle of induction
Bortezomib-based induction Vincristine-based induction
p value
treatment is described as early onset, and after two to
treatment (n=250)
treatment (n=250)
three cycles of induction treatment as late onset.
Baseline peripheral
8 (3%)
13 (5%)
0·37
Vincristine-induced peripheral neuropathy was used as a
neuropathy
reference when we assessed the incidence and severity of
Peripheral neuropathy after one cycle
bortezomib-induced peripheral neuropathy.
Grade 2­4
20 (8%)
11 (4%)
0·27
RNA isolation and microarray processing was done as
Grade 2
10 (50%)
9 (82%)
··
previously described.20 Microarray data presented in this
Grade 3
7 (35%)
1 (9%)
0·18*
report have been stored in the Gene Expression Omnibus
Grade 4
3 (15%)
1 (9%)
··
database (National Center for Biotechnology Information,
Peripheral neuropathy after two or three cycles
Bethesda MD, USA), accession number GSE19784. Gene
Grade 2­4
63 (25%)
17 (7%)
<0·0001
expression arrays were done with RNA extracted
Grade 2
31 (49%)
11 (65%)
··
from myeloma plasma cells that were purified (80%)
Grade 3
24 (38%)
6 (35%)
0·72*
from the extra bone marrow aspiration taken at diagnosis
Grade 4
8 (13%)
0
··
and met the criteria for quality.20
DNA was extracted from peripheral blood nucleated
Data are number (%), unless otherwise indicated. The denominator for calculation of the percentages of patients with
cells or CD138-negative bone marrow cells and quantified
grades 2, 3, and 4 drug-related peripheral neuropathy was the total number of patients presenting with drug-related
by use of the Nanodrop Spectrophotometer (Nanodrop,
peripheral neuropathy after one or two to three cycles of treatment, respectively. *For difference in percentage of
patients with grade 3 and 4 peripheral neuropathy associated with bortezomib and vincristine among the total
Wilmington, DE, USA). Samples were genotyped by use
number of patients presenting with grade 2­4 bortezomib-associated and vincristine-associated peripheral
of the Affymetrix Targeted Genotyping (Affymetrix, Santa
neuropathy, respectively.
Clara, CA, USA) custom built panel, with 3404 SNPs,
Table 1: Incidence of baseline, bortezomib-induced, and vincristine-induced peripheral neuropathy
selected with a hypothesis-driven strategy, targeting
genes and SNPs for which associations or putative
functional effects have been noted (Corthals SL,
This study is registered as an International Standard
unpublished data).
Randomised Controlled Trial, number ISRCTN64455289.
Statistical analysis
Role of the funding source
For differences in incidence of baseline and grade 2­4
The sponsors had no role in the design, gathering,
peripheral neuropathy after one cycle and after two to
analysis, and interpretation of the data, or the writing of
three cycles of bortezomib-based and vincristine-based
the report. The corresponding author had full access to
treatment, ² analysis was done with a two-sided p value
all the data and the final responsibility to submit for
of 0·05. For gene expression data, class comparison of
publication.
groups of arrays was done with one-way ANOVA in Partek
Genomics Suite (version 6.4), followed by multiple-test
Results
correction with a false discovery rate of less than 0·05.
We did gene expression arrays for 329 (39%; 170 treated
For SNP genotyping data, deviations from Fisher's
with bortezomib, 159 treated with vincristine) of
exact t test for Hardy-Weinberg equilibrium at p<0·00001
833 patients included in the trial, and SNP profiles for
and bias in missing data were controlled for each SNP.
samples taken from 369 (44%; 186 treated with
SNPs with a minor allele frequency of less than 5% and
bortezomib, 183 treated with vincristine) patients.
a call rate of less than 80% were removed from further
Simultaneous gene expression and SNP data were
analysis. To assess SNP associations with treatment-
obtained for 185 patients; only SNP data were available
related peripheral neuropathy and calculation of odds
for 184 patients, and only gene expression data were
ratios (ORs), a Cochran-Armitage trend test and a
available for 144 patients. The baseline clinical
Fisher's t exact test were done. We assessed the genomic
characteristics of 513 patients included in this study
inflation factor based on the median ² for each
were not different from the whole patient group included
analysis with PLINK (version 1.07).21 To account for
in the trial (webappendix p 1).
See Online for webappendix
multiple testing, 10 000 permutation tests were done
Table 1 shows the incidence of peripheral neuropathy
with the max(T) permutation procedure with PLINK. To
at baseline and after treatment with bortezomib-based
assess the effect of non-synonymous SNPs associated
and vincristine-based induction treatments in 500 of
with bortezomib-induced peripheral neuropathy and
513 patients who were fully assessable and had a
vincristine-induced peripheral neuropathy, SNPs were
minimum follow-up of 40 months. The median time
characterised by use of the prediction program Sorting
to development of bortezomib-induced peripheral neuro-
Intolerant From Tolerant (version 4.0.3).
pathy was 42 days (range 0­137). Cumulative dose of
Analysis of the gene and SNP sets for peripheral
bortezomib given before development of peripheral
neuropathy associated with bortezomib and vincristine
neuropathy was 13 mg/m². 52 patients (21%) developed
was done by use of Ingenuity Pathway Analysis software
grade 1 bortezomib-induced peripheral neuropathy, and
(version 8.7).
34 (14%) developed grade 1 peripheral neuropathy before
www.thelancet.com/oncology Vol 11 November 2010
1059
Articles
progressing to a higher grade. When patients developed
(range 0­171). Cumulative dose of vincristine given
peripheral neuropathy, the dose of bortezomib was
before development of peripheral neuropathy was 4 mg.
adjusted according to the established guidelines for dose
60
(24%)
patients
developed
vincristine-induced
modification.5­7,19 Median
time
to
development
of
peripheral neuropathy, and 18 (7%) developed grade 1
vincristine-induced peripheral neuropathy was 37 days
peripheral neuropathy before progressing to a higher
Gene name
Gene description
Factor difference
p value
in expression
Grade 2­4 peripheral neuropathy (n=15) versus no peripheral neuropathy (n=134) after one cycle of bortezomib
225189_s_at
RAPH1
Ras association (RalGDS/AF-6) and pleckstrin homology domains 1
2·24
3·04×10²
235014_at
LOC147727
Hypothetical protein LOC147727
2·15
1·91×10²
1569872_a_at
LOC650392
Hypothetical protein LOC650392
1·98
9·65×10
213056_at
FRMD4B
FERM domain containing 4B
1·74
8·42×10³
227984_at
LOC650392
Hypothetical protein LOC650392
1·71
1·19×10³
225478_at
MFHAS1
Malignant fibrous histiocytoma amplified sequence 1
1·68
5·34×10
226913_s_at
SOX8
SRY (sex determining region Y)-box 8
1·68
4·28×10¹³
204810_s_at
CKM
Creatine kinase, muscle
1·67
1·11×10³
1569871_at
LOC650392
Hypothetical protein LOC650392
1·65
1·77×10¹
228057_at
DDIT4L
DNA-damage-inducible transcript 4-like
1·59
5·59×10²
Grade 2­4 peripheral neuropathy (n=44) versus no peripheral neuropathy (n=78) after two or three cycles of bortezomib
205590_at
RASGRP1
RAS guanyl releasing protein 1 (calcium and DAG regulated)
2·97
2·14×10²
204527_at
MYO5A
Myosin VA (heavy chain 12, myoxin)
1·93
3·21×10²
235065_at
··
··
1·57
3·19×10²
205422_s_at
ITGBL1
Integrin, -like 1 (with EGF-like repeat domains)
1·44
1·35×10³
228113_at
RAB37
RAB37, member of RAS oncogene family
1·41
3·69×10²
210321_at
GZMH
Granzyme H (cathepsin G-like 2, protein h-CCPX)
1·37
3·19×10²
226969_at
MTR
5-methyltetrahydrofolate-homocysteine methyltransferase
1·34
4·26×10²
204072_s_at
FRY
Furry homolog (Drosophila)
1·31
4·94×10²
236442_at
DPF3
D4, zinc and double PHD fingers, family 3
1·30
3·38×10²
243329_at
··
··
1·30
4·26×10²
Grade 2­4 peripheral neuropathy (n=9) versus no peripheral neuropathy (n=129) after one cycle of vincristine
208235_x_at
GAGE7
G antigen 7
11·55
3·21×10³
206640_x_at
GAGE12I
G antigen 12I
11·46
4·29×10³
207739_s_at
GAGE2C
G antigen 2C
7·76
1·62×10³
208155_x_at
GAGE6
G antigen 6
6·88
1·06×10
206897_at
PAGE1
P antigen family, member 1 (prostate associated)
6·76
4·29×10²
216063_at
HBBP1
Haemoglobin, pseudogene 1
6·24
4·04×10²
207086_x_at
GAGE4
G antigen 4
6·16
3·29×10
206626_x_at
SSX1
Synovial sarcoma, X breakpoint 1
5·93
2·61×10²
207912_s_at
DAZ1
Deleted in azoospermia 1
5·86
1·06×10
214957_at
ACTL8
Actin-like 8
4·93
1·32×10¹
Grade 2­3 peripheral neuropathy (n=10) versus no peripheral neuropathy (n=103) after two or three cycles of vincristine
210632_s_at
SGCA
Sarcoglycan, alpha (50 kDa dystrophin-associated glycoprotein)
4·08
3·35×10²
210992_x_at
FCGR2C
Fc fragment of IgG, low affinity IIc, receptor for (CD32)
2·49
3·57×10²
241991_at
··
··
1·80
3·35×10²
206771_at
UPK3A
Uroplakin 3A
1·59
2·94×10²
241365_at
··
··
1·57
3·35×10²
236266_at
RORA
RAR-related orphan receptor A
1·53
2·94×10²
214059_at
IFI44
Interferon-induced protein 44
1·51
4·92×10
230477_at
··
··
1·48
2·94×10²
237322_at
MIAT
Myocardial infarction associated transcript (non-protein coding)
1·45
2·94×10²
239239_at
··
··
1·33
2·94×10²
First column is the probe-set identification numbers. Genes were ranked from highest to lowest change; the first ten genes with the highest changes are shown.
Table 2: Differentially expressed genes in early-onset and late-onset bortezomib-induced and vincristine-induced peripheral neuropathy
1060
www.thelancet.com/oncology Vol 11 November 2010
Articles
grade. When patients developed vincristine-induced
sets are shown in the webappendix pp 2­3. The genes
peripheral neuropathy, vincristine was discontinued and
showing the highest change in expression included RAPH1
supportive treatments such as pregabalin were used.
(involved in signal transduction), FRMD4B, MFHAS1
Overall, baseline peripheral neuropathy was noted in
(possibly an oncogene regulated by NFB or tumour
only a small number of patients (table 1). The proportion
necrosis factor), and DDIT4L (a DNA-damage inducible
of patients developing late-onset bortezomib-induced
transcript; table 2). Genes that might play a direct part in
peripheral neuropathy was significantly higher than that
bortezomib-induced
peripheral
neuropathy
are
of patients with late-onset vincristine-induced peripheral
transcription regulator SOX8 (involved in development of
neuropathy (table 1).
peripheral nervous system), and CPT1C and RHOBTB2
Gene expression arrays for 15 patients developing early-
(webappendix pp 2­3). Ingenuity pathway analyses of gene
onset grade 2­4 bortezomib-induced peripheral neuropathy
and SNP sets showed enrichment of genes implicated in
were compared with arrays of patients who did not develop
the canonical pathway of signalling mediated by AMP-
bortezomib-induced peripheral neuropathy (table 2). Grade
activated protein kinase (AMPK), including CPT1C, CKM,
2­4 early-onset bortezomib-induced peripheral neuropathy
and PIK3CG (three of 156 genes involved in AMPK
was characterised by 19 differentially expressed genes (false
signalling were upregulated, p=7·33×10­).
discovery rate <0·05). The genes showing the highest
Gene expression arrays for 44 patients with grade 2­4
changes in the gene expression arrays are shown in table 2,
late-onset bortezomib-induced peripheral neuropathy
and the complete number of differentially expressed probe
were characterised by 27 differentially expressed genes,
Chromosome
Gene
Single-nucleotide polymorphism type
Odds ratio (95% CI)
p value
Permuted p value
Grade 2­4 peripheral neuropathy (n=13) versus no peripheral neuropathy (n=147) after one cycle of bortezomib
rs2251660
17
RDM1
Coding non-synonymous
3·65 (1·55­8·57)
9·06×10
2·40×10³
rs4646091
1
CASP9
Intron
3·59 (1·59­8·14)
1·43×10³
2·90×10³
rs1126667
17
ALOX12
Coding non-synonymous
3·50 (1·47­8·32)
2·95×10³
3·80×10³
rs434473
17
ALOX12
Coding non-synonymous
3·50 (1·47­8·32)
2·95×10³
4·10×10³
rs7823144
8
LSM1
Intron
4·11 (1·48­11·39)
2·30×10³
7·60×10³
rs1879612
15
IGF1R
Intron
0·22 (0·07­0·77)
9·42×10³
8·30×10³
rs1029871
3
NEK4
Coding non-synonymous
0·30 (0·11­0·81)
8·31×10³
9·30×10³
Grade 2­4 peripheral neuropathy versus (n=49) no peripheral neuropathy (n=80) after two or three cycles of bortezomib
rs1799800
16
ERCC4
Intron
2·74 (1·56­4·84)
5·16×10
1·00×10³
rs1799801
16
ERCC4
Coding synonymous
2·48 (1·43­4·28)
8·85×10
1·10×10³
rs2300697
2
SRD5A2
Intron
0·63 (0·37­1·05)
4·80×10²
2·90×10³
rs1059293
21
IFNGR2
Untranslated, intron
2·30 (1·37­3·87)
8·97×10
3·20×10³
rs2276583
2
ERCC3
Locus
1·26 (0·75­2·12)
3·87×10¹
3·30×10³
rs189037
11
ATM
Locus, untranslated
0·53 (0·32­0·89)
2·32×10²
3·60×10³
rs10501815
11
MRE11A
Intron, TagSNP
3·27 (1·39­7·74)
4·41×10³
4·20×10³
rs664677
11
ATM
Intron
0·57 (0·34­0·96)
4·36×10²
5·90×10³
rs664982
11
ATM
Intron
0·51 (0·30­0·85)
1·72×10²
6·20×10³
rs6131
1
SELP
Coding non-synonymous
0·43 (0·23­0·83)
6·69×10³
6·30×10³
rs1130499
7
PTPRN2
Coding non-synonymous
0·43 (0·23­0·79)
6·23×10³
6·60×10³
rs4722266
7
STK31
Coding non-synonymous
0·29 (0·12­0·74)
5·66×10³
8·30×10³
rs2267668
6
PPARD
Intron
0·35 (0·15­0·83)
9·30×10³
9·10×10³
Grade 2­4 peripheral neuropathy versus (n=7) no peripheral neuropathy (n=151) after one cycle of vincristine
rs7739752
6
PPARD
Intron
13·43 (3·90­46·22)
6·34×10
8·00×10
rs2288087
9
ALDH1A1
Intron, TagSNP
7·62 (1·68­34·65)
1·40×10³
1·50×10³
rs1494961
4
HEL308
Coding non-synonymous
6·67 (1·47­30·32)
2·30×10³
2·60×10³
rs6901410
6
PPARD
Intron
9·67 (2·65­35·30)
7·75×10
6·00×10³
rs6902123
6
PPARD
Intron
9·67 (2·65­35·30)
7·75×10
6·00×10³
rs2274407
13
ABCC4
Coding non-synonymous
7·15 (2·02­25·31)
2·94×10
6·10×10³
rs909253
6
LTA
Intron
4·67 (1·52­14·34)
3·09×10³
6·60×10³
rs6457816
6
PPARD
Intron
8·89 (2·46­32·17)
1·40×10
7·30×10³
rs1041981
6
LTA
Coding non-synonymous
4·52 (1·47­13·88)
3·58×10³
7·40×10³
rs3803258
13
SLC10A2
Untranslated
4·30 (1·45­12·74)
3·51×10³
7·40×10³
rs3749442
3
ABCC5
Coding synonymous
4·64 (1·5 ­14·05)
2·72×10³
9·60×10³
(Continued on next page)
www.thelancet.com/oncology Vol 11 November 2010
1061
Articles
Chromosome
Gene
Single-nucleotide polymorphism type
Odds ratio (95% CI)
p value
Permuted p value
(Continued from previous page)
Grade 2­3 peripheral neuropathy (n=14) versus no peripheral neuropathy (n=104) after two or three cycles of vincristine
rs10515114
5
CART
Locus
4·62 (1·68­12·72)
7·92×10
2·90×10³
rs6873545
5
GHR
Intron
0·09 (0·01­0·69)
3·44×10³
3·60×10³
rs3734354
6
SIM1
Coding non-synonymous
3·30 (1·39­7·82)
2·31×10³
5·10×10³
rs11688
1
JUN
Coding synonymous
5·00 (1·80­13·91)
9·10×10
5·20×10³
rs4129472
5
GHR
Intron
0·11 (0·01­0·80)
6·46×10³
5·20×10³
rs1413239
1
DPYD
Intron, TagSNP
3·29 (1·47­7·37)
3·03×10³
5·40×10³
rs1045020
5
SLC22A5
Untranslated
4·80 (1·83­12·61)
1·48×10³
5·40×10³
rs9885672
6
KIAA0274
Coding non-synonymous
3·89 (1·62­9·33)
2·05×10³
5·60×10³
rs3887412
16
ABCC1
Intron, TagSNP
3·36 (1·47­7·67)
3·31×10³
5·70×10³
rs6886047
5
GHR
Intron
0·10 (0·01­0·72)
3·97×10³
6·10×10³
rs1236913
9
PTGS1
Coding non-synonymous
5·40 (1·79­16·28)
1·43×10³
6·30×10³
rs2644983
16
ABCC1
Intron, TagSNP
4·22 (1·69­10·50)
2·27×10³
6·60×10³
rs1042713
5
ADRB2
Coding non-synonymous
0·23 (0·08­0·69)
5·30×10³
7·20×10³
rs1966265
5
FGFR4
Coding non-synonymous
3·47 (1·51­7·94)
3·40×10³
7·30×10³
rs2308327
10
MGMT
Coding non-synonymous
3·38 (1·33­8·58)
3·69×10³
7·30×10³
rs5759197
22
BZRP
Intron
2·93 (1·31­6·53)
6·32×10³
7·60×10³
rs1005658
22
BZRP
Locus
3·14 (1·39­7·08)
6·04×10³
8·50×10³
rs7441774
4
UGT2B7
Intron
3·60 (1·40­9·23)
6·61×10³
9·60×10³
Table 3: Single-nucleotide polymorphisms associated with bortezomib-induced and vincristine-induced peripheral neuropathy
using the same false discovery rate as for early onset
associated SNPs located in genes involved in the
(webappendix pp 2­3), and showed a different pattern of
development and function of the nervous system (three
gene expression to that in early-onset bortezomib-
genes, p=3·35×10­³­1·69×10­²) and in inflammatory
induced peripheral neuropathy, without overlap (table 2).
disease (26 genes, p=2·09×10­³­4·95×10­³).
RASGRP1 showed the highest change in patients with
The genetic profile of myeloma plasma cells from nine
late-onset bortezomib-induced peripheral neuropathy
patients who developed grade 2­4 early-onset vincristine-
compared with patients without this side-effect (table 2).
induced peripheral neuropathy showed overexpression
Furthermore, we noted upregulation of genes involved
of the genes for testis cancer antigens, of which the
in transcription regulation, including TRERF1, TRPS1,
GAGE genes were mainly upregulated (table 2).
and MDM2. We noted enrichment of genes involved in
The gene profiles of ten patients who developed grade
the development and function of the nervous system,
2 or 3 late-onset vincristine-induced peripheral neuro-
including SOD2 and MYO5A.
pathy showed only ten differentially expressed genes,
All significant SNPs (permuted p<0·01) associated
including RORA and IFI44 (table 2).
with
grade
2­4
early-onset
bortezomib-induced
Table 3 shows SNPs significantly associated with early-
peripheral neuropathy are shown in table 3 (values of
onset vincristine-induced peripheral neuropathy. Four of
permuted p<0·05 are shown in webappendix pp 4­10).
the most highly associated SNPs (rs7739752, rs6901410,
Several SNPs associated with early-onset bortezomib-
rs6902123, and rs6457816) were located in the transcription
induced peripheral neuropathy were located in caspase 9
factor PPARD. Additionally, an intronic (rs909253) and a
(rs4646091, rs2020895, rs2020903, rs4646032, and
coding non-synonymous SNP (rs1041981) in LTA were
rs4646034). Other highly associated SNPs were located
significantly associated with early-onset vincristine-
in genes RDM1, ALOX12, IGF1R, and LSM1 (table 3).
induced peripheral neuropathy. Other significant SNPs
Pathway analysis of these associated genes showed
were located in genes for transporter enzymes ABCC4,
enrichment of genes involved in cell death (14 genes,
ABCC5, and SLC10A2, oxidising enzyme ALDH1A1, and
p=5·25×10­³­4·93×10­²),
DNA
repair
(14
genes,
GLI1I (table 3; webappendix pp 4­10). Pathway analysis
p=5·25×10­³­4·93×10­²), and development and function
showed enrichment of associated SNPs located in genes
of the nervous system (four genes, p=2·01×10­³).
involved in cellular growth and proliferation (four genes,
The SNPs that were characteristic of late-onset
p=1·14×10­²­4·95×10­²).
bortezomib-induced peripheral neuropathy were mainly
Some intronic SNPs in the dihydropyrimidine
located in DNA repair genes, such as ERCC3, ERCC4,
dehydrogenase gene DPYD and some in the ABC
ATM, BRCA1, EXO1, and MRE11A (table 3; webappendix
transporter gene ABCC1 were associated with late-onset
pp 4­10). Pathway analysis showed enrichment of
vincristine-induced peripheral neuropathy (table 3).
1062
www.thelancet.com/oncology Vol 11 November 2010
Articles
Pathway analysis showed that most significant SNPs
peripheral neuropathy were located in SERPINB2
(permuted p<0·05) were located in genes for absorption,
(plasminogen activator inhibitor-2). SERPINB2, with
distribution, metabolism, and excretion (six genes,
SERPIN-1 (plasminogen activator inhibitor-1), tissue-type
p=2·06×10­²­4·18×10­²).
plasminogen activator, and urokinase-type plasminogen
activator, has been shown to be induced in dorsal root
Discussion
ganglion neurons after peripheral axotomy in mice.26
The genetic profiles of patients with early-onset
These serpins might also act as autocrine or paracrine
bortezomib-induced peripheral neuropathy suggest the
regulators of plasminogen-activator-mediated nerve
involvement of genes involved in transcription, apoptosis,
regeneration processes.26 The associated SNPs might
and AMPK-mediated signalling. The possible role of
affect SERPINB2 expression through their effect on
AMPK-mediated signalling is of particular interest
splicing
regulation.
Besides
genes
involved
in
because this enzyme functions by stimulating the
development of the nervous system, proinflammatory
signalling pathways that replenish cellular ATP supplies
genes might play an important part in the pathogenesis
in response to low glucose, hypoxia, ischaemia, or heat
of late-onset bortezomib-induced peripheral neuropathy,
shock, which might be triggered in myeloma cells in
based on the presence of intronic SNPs in MBL2 and
response to bortezomib. CPT1C codes for an enzyme
PPARD (Corthals SL, unpublished data), and of about
found in neuron mitochondria that is involved in
30% of SNPs with reported inflammatory roles. The
transport of hydrophobic fatty acid chains into
hypothesis that the DNA repair pathway is involved in
mitochondria, and plays a part in mitochondrial
bortezomib-induced peripheral neuropathy, and that this
dysfunction. It might also have an important role in
side-effect might be caused by the inability to repair
bortezomib-induced
peripheral
neuropathy,
since
neuronal damage (Corthals SL, unpublished data), could
damage to mitochondria and endoplasmic reticulum
be substantiated by the presence of SNPs in BRCA1
through activation of a mitochondrial-based apoptotic
(rs16941 and rs799917). These non-synonymous SNPs
pathway by bortezomib was noted in dorsal root ganglia
might have an effect on the phosphorylation state of a
of mice given bortezomib.11 RHOBTB2, encodes another
protein, which has been shown to abolish the P871L
enzyme implicated here, has been shown to be
phosphorylation site in BRCA1.27 Therefore, early-onset
upregulated during drug-induced apoptosis, being
and
late-onset
bortezomib-induced
peripheral
mainly dependent on E2F1.22 Knockout of RHOBTB2
neuropathies were both associated with a myeloma
with small interfering RNAs has been shown to delay the
genetic profile that was characterised by genes involved
onset of drug-induced apoptosis.22 RASGRP1 is involved
in the development of the nervous system; however,
in many processes, including apoptosis and calcium-ion
apoptosis was also a characteristic for the development of
binding, which are potentially interesting for its role in
early-onset bortezomib-induced peripheral neuropathy.
bortezomib-induced peripheral neuropathy. The presence
Genetic polymorphisms in genes involved in nervous
of polymorphisms in the apoptosis gene caspase 9, which
system development and DNA repair play a part in both
plays an important part in bortezomib-induced apoptosis,
the early and late onset of this side-effect.
suggests the possible contribution of this enzyme to
A comparison of the molecular profiles of bortezomib-
early-onset peripheral neuropathy.23,24 One of the most
induced peripheral neuropathy and vincristine-induced
significant SNPs (rs1029871) might have a role in the
peripheral neuropathy showed no overlap in associated
splicing regulation of NEK4, which is involved in the
genes or SNPs. Genes involved in cell cycle and
regulation of cell cycle and cell division. Furthermore,
proliferation were mainly associated with early-onset
SNPs in enriched pathways like DNA repair and nervous
vincristine-induced peripheral neuropathy, both in the
system development and function were associated with
analyses of genetic pathways and SNPs. Additionally,
early-onset bortezomib-induced peripheral neuropathy.
involvement of proinflammatory genes in early-onset
Late-onset bortezomib-induced peripheral neuropathy
vincristine-induced
peripheral
neuropathy
was
was associated with genes involved in the development
substantiated by the finding of SNPs in PARP1 and LTA,
and function of the nervous system. We noted
and two SNPs in GLI1 (rs2228224 and rs2228226), which
upregulation of the superoxide dismutase gene SOD2 in
both encode an aminoacid change; rs2228226 has been
myeloma plasma cells; SOD2 is regulated by tumour
shown to affect GLI1 activity, thereby affecting the
necrosis factor and NFB, and is known to have a role
inflammatory response.28
in the survival of neurons. Patients with diabetes and a
Genes implicated in drug absorption, distribution,
polymorphism in the SOD2 gene, leading to reduced
metabolism, and excretion have been shown to be
SOD2 activity, have been shown to be at increased risk of
involved in chemotherapy-induced peripheral neuro-
developing
diabetic
peripheral
neuropathy.25
The
pathy.29 In accordance with this finding, an association
protective effect of SOD2 might be eliminated with
was noted for late-onset vincristine-induced peripheral
bortezomib-induced apoptosis, which might trigger a
neuropathy with nine intronic SNPs in ABCC1;
susceptibility to oxidative stress in treated patients. Three
vincristine is known to be a substrate of the protein coded
SNPs associated with late-onset bortezomib-induced
for by this gene.
www.thelancet.com/oncology Vol 11 November 2010
1063
Articles
International Myeloma Foundation (unrestricted grant), Erasmus MC
Panel: Research in context
(translational research grant), and Janssen-Cilag Orthobiotech
(unrestricted research grant). We thank participants of the
Systematic review
HOVON-65/GMMG-HD4 trial.
We did a systematic review of chemotherapy-induced
References
peripheral neuropathy, particularly bortezomib-induced
1
Harousseau JL, Attal M, Leleu X, et al. Bortezomib plus
dexamethasone as induction treatment prior to autologous stem
peripheral neuropathy, before doing the molecular analysis
cell transplantation in patients with newly diagnosed multiple
in our study.4
myeloma: results of an IFM phase II study.
Haematologica 2006; 91: 1498­505.
Interpretation
2
Rosinol L, Oriol A, Mateos MV, et al. Phase II PETHEMA trial
Our data confirm the role of myeloma-related factors in
of alternating bortezomib and dexamethasone as induction
regimen before autologous stem-cell transplantation in younger
bortezomib-induced peripheral neuropathy16 and indicate
patients with multiple myeloma: efficacy and clinical
that both the inherited genetic constitution of the host
implications of tumor response kinetics. J Clin Oncol 2007;
25: 4452­58.
(patient) and the tumour (myeloma) should be thought of
3
Sonneveld P, van der Holt B, Schmidt-Wolf IGH, et al. First
as contributors to the risk of bortezomib-induced
analysis of HOVON-65/GMMG-HD4 randomized phase III trial
peripheral neuropathy.
comparing bortezomib, adriamycine, dexamethasone (PAD) vs VAD
as induction treatment prior to high dose melphalan (HDM) in
patients with newly diagnosed multiple myeloma (MM). Blood 2008;
In conclusion, this study provides the first large dataset
112: 653.
in which the contribution of both the inherited genetic
4
Delforge M, Bladé J, Dimopoulos MA, et al. Treatment-related
constitution of the host (patient) and the tumour
peripheral neuropathy in multiple myeloma: the challenge
continues. Lancet Oncol (in press).
(myeloma) to the development of bortezomib-induced
5
Jagannath S, Barlogie B, Berenson J, et al. A phase 2 study of
peripheral neuropathy has been reported (panel). We
two doses of bortezomib in relapsed or refractory myeloma.
identified molecular factors that are associated with
Br J Haematol 2004; 127: 165­72.
bortezomib-induced peripheral neuropathy in patients
6
Richardson PG, Barlogie B, Berenson J, et al. A phase 2 study of
bortezomib in relapsed, refractory myeloma. N Engl J Med 2003;
with newly diagnosed multiple myeloma. Genes for
348: 2609­17.
apoptosis contribute to early-onset bortezomib-induced
7
Richardson PG, Briemberg H, Jagannath S, et al. Frequency,
peripheral neuropathy, whereas genes that have a role in
characteristics, and reversibility of peripheral neuropathy during
treatment of advanced multiple myeloma with bortezomib.
inflammatory pathways and DNA repair contribute to
J Clin Oncol 2006; 24: 3113­20.
the development of late-onset peripheral neuropathy,
8
Mateos MV. Management of treatment-related adverse events
indicating that distinct genetic factors are involved in the
in patients with multiple myeloma. Cancer Treat Rev 2010;
36 (suppl 2): S24­32.
development of early-onset and late-onset forms of this
9
Badros A, Goloubeva O, Dalal JS, et al. Neurotoxicity of bortezomib
side-effect. Bortezomib-induced and vincristine-induced
therapy in multiple myeloma: a single-center experience and review
peripheral neuropathy arise through different molecular
of the literature. Cancer 2007; 110: 1042­49.
mechanisms.
Our
findings
strongly
suggest
an
10
Cata JP, Weng HR, Burton AW, Villareal H, Giralt S, Dougherty PM.
Quantitative sensory findings in patients with bortezomib-induced
interaction between myeloma-related factors and the
pain. J Pain 2007; 8: 296­306.
patient's genetic background in the development of
11
Cavaletti G, Gilardini A, Canta A, et al. Bortezomib-induced
bortezomib-induced peripheral neuropathy. Profiles
peripheral neurotoxicity: a neurophysiological and pathological
study in the rat. Exp Neurol 2007; 204: 317­25.
of genetic risk might be used in future to identify
12
Landowski TH, Megli CJ, Nullmeyer KD, Lynch RM, Dorr RT.
patients with an increased risk of bortezomib-induced
Mitochondrial-mediated disregulation of Ca2+ is a critical
peripheral neuropathy.
determinant of Velcade (PS-341/bortezomib) cytotoxicity in
myeloma cell lines. Cancer Res 2005; 65: 3828­36.
Contributors
13
Ravaglia S, Corso A, Piccolo G, et al. Immune-mediated
AB designed and participated in the research, analysed data, and wrote
neuropathies in myeloma patients treated with
the report. SLC participated in the research, analysed data, and wrote the
bortezomib. Clin Neurophysiol 2008; 119: 2507­12.
report. JLMJ is a neurologist who reviewed the report. BVDH was the
14
Ropper AH, Gorson KC. Neuropathies associated with
statistician for the trial, gathered data, and reviewed the report. RK and
paraproteinemia. N Engl J Med 1998; 338: 1601­07.
MVD analysed data and reviewed the report. YDK participated in the
15
Borrello I, Ferguson A, Huff CA, et al. Bortezomib and
research and reviewed the report. LEJ and UB were the trial managers,
thalidomide treatment of newly diagnosed patients with
and reviewed the report. HML was the study coordinator and reviewed
multiple myeloma--efficacy and neurotoxicity. Blood 2006;
the report. BGD designed the research and reviewed the report. HG was
108: 3528 (abstr).
the study coordinator, and designed the research and reviewed the
16
Richardson PG, Xie W, Mitsiades C, et al. Single-agent
report. PS was the principal investigator of the study, designed the
bortezomib in previously untreated multiple myeloma: efficacy,
research, and reviewed the report.
characterization of peripheral neuropathy, and molecular
correlations with response and neuropathy. J Clin Oncol 2009;
Conflicts of interest
27: 3518­25.
PS has served on advisory boards for Johnson & Johnson and
17
Roccaro AM, Vacca A, Ribatti D. Bortezomib in the treatment
Millennium Pharmaceuticals. HG has served on advisory boards
of cancer. Recent Pat Anticancer Drug Discov 2006; 1: 397­403.
for Johnson & Johnson. The other authors declare that they have no
18
Trotti A, Colevas AD, Setser A, et al. CTCAE v3.0: development
conflicts of interest.
of a comprehensive grading system for the adverse effects of cancer
treatment. Semin Radiat Oncol 2003; 13: 176­81.
Acknowledgments
19
Richardson PG, Sonneveld P, Schuster MW, et al. Bortezomib
This work was funded by the German Federal Ministry of Education
or high-dose dexamethasone for relapsed multiple myeloma.
and Research, Dutch Cancer Foundation Queen Wilhelmina,
N Engl J Med 2005; 352: 2487­98.
European Hematology Association (clinical research grant),
1064
www.thelancet.com/oncology Vol 11 November 2010
Articles
20
Broyl A, Hose D, Lokhorst H, et al. Gene expression profiling for
25
Strokov IA, Bursa TR, Drepa OI, Zotova EV, Nosikov VV,
molecular classification of multiple myeloma in newly diagnosed
Ametov AS. Predisposing genetic factors for diabetic
patients. Blood 2010; published online June 23.
polyneuropathy in patients with type 1 diabetes: a population-based
DOI:10.1182/blood-2009-12-261032.
case-control study. Acta Diabetol 2003; 40 (suppl 2): S375­79.
21
Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for
26
Yamanaka H, Obata K, Fukuoka T, et al. Induction of plasminogen
whole-genome association and population-based linkage analyses.
activator inhibitor-1 and -2 in dorsal root ganglion neurons after
Am J Hum Genet 2007; 81: 559­75.
peripheral nerve injury. Neuroscience 2005; 132: 183­91.
22
Freeman SN, Ma Y, Cress WD. RhoBTB2 (DBC2) is a mitotic E2F1
27
Savas S, Ozcelik H. Phosphorylation states of cell cycle and DNA
target gene with a novel role in apoptosis. J Biol Chem 2008;
repair proteins can be altered by the nsSNPs. BMC Cancer 2005;
283: 2353­62.
5: 107.
23
Voortman J, Checinska A, Giaccone G, Rodriguez JA, Kruyt FA.
28
Lees CW, Zacharias WJ, Tremelling M, et al. Analysis of germline
Bortezomib, but not cisplatin, induces mitochondria-dependent
GLI1 variation implicates hedgehog signalling in the regulation of
apoptosis accompanied by up-regulation of noxa in the non-small
intestinal inflammatory pathways. PLoS Med 2008; 5: e239.
cell lung cancer cell line NCI-H460. Mol Cancer Ther 2007;
29
Johnson DC, Ramos C, Szubert AJ, et al. Genetic variation in ADME
6: 1046­53.
genes is associated with thalidomide related peripheral neuropathy
24
Poulaki V, Mitsiades CS, Kotoula V, et al. The proteasome inhibitor
in multiple myeloma patients. Blood 2008; 112: 1675 (abstr).
bortezomib induces apoptosis in human retinoblastoma cell lines
in vitro. Invest Ophthalmol Vis Sci 2007; 48: 4706­19.
www.thelancet.com/oncology Vol 11 November 2010
1065

Document Outline