significance of tp53 mutations determined by next-generation “deep” sequencing in prognosis of...

Cancer Letters xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Cancer Letters

journal homepage: www.elsevier .com/locate /canlet

Significance of TP53 mutations determined by next-generation ‘‘deep’’sequencing in prognosis of estrogen receptor-positive breast cancer

0304-3835/$ - see front matter � 2013 Elsevier Ireland Ltd. All rights reserved.http://dx.doi.org/10.1016/j.canlet.2013.08.028

⇑ Corresponding author. Address: Department of Breast and Endocrine Surgery,Osaka University Graduate School of Medicine, 2-2-E10 Yamadaoka, Suita-shi,Osaka 565-0871, Japan. Tel.: +81 6 6879 3772; fax: +81 6 6879 3779.

E-mail address: [email protected] (S. Noguchi).

Please cite this article in press as: K. Uji et al., Significance of TP53 mutations determined by next-generation ‘‘deep’’ sequencing in prognosis of ereceptor-positive breast cancer, Cancer Lett. (2013), http://dx.doi.org/10.1016/j.canlet.2013.08.028

Kumiko Uji, Yasuto Naoi, Naofumi Kagara, Masafumi Shimoda, Atsushi Shimomura, Naomi Maruyama,Kenzo Shimazu, Seung Jin Kim, Shinzaburo Noguchi ⇑Department of Breast and Endocrine Surgery, Osaka University Graduate School of Medicine, Osaka, Japan

a r t i c l e i n f o

Article history:Received 31 May 2013Received in revised form 31 July 2013Accepted 15 August 2013Available online xxxx

Keywords:Breast cancerTP53 mutationNext-generation deep sequencing

a b s t r a c t

Next-generation ‘‘deep’’ sequencing (NGS) was used for the mutational analysis of TP53, and a DNAmicroarray was used for the determination of the TP53 mutation-associated gene expression signature(TP53 GES) in 115 estrogen receptor (ER)-positive breast cancers. NGS detected 27 TP53 mutations, ofwhich 20 were also detected by Sanger sequencing (SS) and seven were detected only by NGS. A signif-icantly higher number of mutant alleles (33.9%) was detected in the tumors with TP53 mutationsdetected by SS compared with those with TP53 mutations detected only by NGS (11.1%). The TP53 muta-tions detected by NGS were more significantly associated with a large tumor size, a high histologicalgrade, progesterone receptor-negativity, and HER2-positivity compared with those detected by SS. TheTP53 mutations detected by SS, but not those detected by NGS or the p53 immunohistochemistry, exhib-ited a significant association with poor prognosis. In addition, the TP53 GES more clearly differentiatedlow- from high-risk patients for relapse than the TP53 mutations detected by SS, regardless of the otherconventional prognostic factors. Thus, NGS is more sensitive for the detection of TP53 mutations, but theprognostic significance of these mutations could not be demonstrated. In contrast, the TP53 GES proved tobe a powerful prognostic indicator for ER-positive tumors.

� 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The tumor suppressor gene TP53 is a transcription factor thatcontrols the expression of a variety of genes that are implicatedin the regulation of the cell cycle, apoptosis, and genomic integrity[1,2]. In addition, this gene has been shown to play an importantrole in the pathogenesis of various cancers, including breast cancer[3]. It has been reported that 15–71% of breast tumors harbor aTP53 mutation and that those tumors with a TP53 mutation havea more aggressive phenotype, such as estrogen receptor(ER)-negativity and high histological grade, compared with thosewithout a TP53 mutation [4,5]. The correlation between the TP53status and prognosis has been studied by many investigators. Mostof these studies indicate that a TP53 mutation is associated withpoor prognosis [6], although contradictory results have also beenreported, especially for patients treated with chemotherapy [7,8]most likely because the chemo-sensitivity of breast tumors is af-fected by the presence of a TP53 mutation [9–11]. In contrast,

rather consistent results have been reported for the associationbetween the presence of a TP53 mutation and poor prognosis inpatients treated with hormonal therapy [12,13].

Breast cancer tissue is composed of a variety of constituents,including not only tumor cells but also stromal fibroblasts andinflammatory cells, and the proportion of tumor cells in tumor tis-sue varies widely from 20% to 95% [14]. The TP53 status is usuallyassessed through the Sanger sequencing (SS) method using DNAsamples extracted from the whole tumor tissue. Because the detec-tion sensitivity of SS is approximately 20% [15], it is speculated thata significant proportion of TP53 mutations could be missed due to alow proportion of tumor cells in the tumor tissue.

The advent of next-generation sequencing (NGS) technologyhas resulted in high-throughput ‘‘deep’’ sequencing with highersensitivity than that achieved with the conventional SS method.In fact, it has been reported that the use of NGS resulted in thedetection of more EGFR mutations in lung cancers than those ob-tained with SS [16,17]. It is therefore tempting to use NGS for thestudy of TP53 mutations in breast tumors because a more accurateassessment of TP53 mutations might result in a further elucidationof the clinicopathological characteristics of breast tumors with amutation in the TP53 gene. However, such a study has yet to be re-ported. Thus, the first aim of the present study was to use NGS for

strogen

http://dx.doi.org/10.1016/j.canlet.2013.08.028

mailto:[email protected]


http://www.sciencedirect.com/science/journal/03043835

http://www.elsevier.com/locate/canlet


2 K. Uji et al. / Cancer Letters xxx (2013) xxx–xxx

an analysis of TP53 mutations to determine whether additionalmutations can be identified with NGS and to gain a better under-standing of the clinicopathological characteristics of breast tumorswith TP53 mutations.

The second objective was to evaluate the prognostic signifi-cance of the TP53 gene expression signature (GES) that was re-ported by Miller et al. [18] and Takahashi et al. [19]. The TP53GES was developed to distinguish TP53-wild-type and -mutantbreast cancers using the signature of the genes that are differen-tially expressed between TP53-wild-type and -mutant breastcancers. These researchers found that the prognostic capability ofthe TP53 GES is superior to that of the TP53 mutation status.However, it is speculated that this superiority stems from, at leastin part, the fact that the TP53 mutations are mostly analyzedthrough SS, which is suspected of overlooking a significant propor-tion of mutations in tumor tissues with a low tumor cell content. Itis thus interesting to compare the prognostic significance of theTP53 GES with that of TP53 mutations determined by NGS, whichhas been shown to be more sensitive than SS. Therefore, thesetwo issues were investigated in the present study using ER-positivebreast tumors treated with adjuvant hormonal therapy for tworeasons. First, the prognostic significance of the TP53 mutationsdetected by SS has been quite consistently demonstrated withoutthe confounding effects of adjuvant chemotherapy. The secondreason is that prognostic classification is clinically very importantin the decision-making process pertaining to the indication ofadjuvant chemotherapy for patients with ER-positive breasttumors.

2. Materials and methods

2.1. Patients and tumor samples

The study cohort consisted of 115 female patients with ER-positive invasivebreast cancer who underwent mastectomy or breast-conserving surgery and subse-quent radiation therapy and were treated with adjuvant hormonal therapy alone atour hospital between 1998 and 2008. The median follow-up for these patients was117 months with a range of 21–173 months. Of these 115 patients, 73 were treatedpostoperatively with tamoxifen (20 mg/day) or toremifene (40 mg/day), 31 weretreated with goserelin (3.6 mg/4 weeks) plus tamoxifen (20 mg/day), one was trea-ted with goserelin (3.6 mg/4 weeks) alone, and 10 were treated with an aromataseinhibitor (1 mg/day anastrozole or 25 mg/day exemestane). Tamoxifen, toremifene,and the aromatase inhibitors were administered for approximately 5 years or untilrecurrence if it occurred earlier, whereas goserelin was administered for approxi-mately 2 years. A total of 32 patients developed recurrences (25 distant and 7 loco-regional recurrences). The tumor tissues obtained at surgery were snap frozen inliquid nitrogen and maintained at �80 �C until use for DNA and RNA extraction,and the surgical specimens were fixed in 10% buffered formaldehyde for histologi-cal analysis. Informed consent for the study was obtained from each patient beforesurgery.

2.2. DNA extraction and Sanger sequencing (SS)

DNA was extracted from the tumor tissues using the DNeasy Blood and TissueKit� (QIAGEN, Germantown, MD, USA) according to the manufacturer’s instruc-tions. The SS analysis of the entire coding region of TP53 (exon 2–11) was conductedusing the Applied Biosystems 3730 DNA analyzer (Applied Biosystems, Foster City,CA, USA), as previously described [20]. The Variant Reporter� Software (version 1.0;Applied Biosystems) was used for the TP53 mutation search.

2.3. Next-generation sequencing (NGS)

The GS Junior system (Roche Diagnostics, Basel, Switzerland) was used for theNGS following the modified protocol described below. (1) Amplicon preparation:the purified PCR products used for SS were also used for NGS. The amplicons of eachexon derived from an individual’s DNA sample were mixed in a length-weightedequal-volume ratio. Then, 500 ng of each mixture was concentrated with Ethachin-mate (Nippon Gene, Tokyo, Japan) and eluted in 16 ll of TE buffer. (2) Ampliconsequencing: the sequence library was prepared in accordance with the Rapid LibraryPreparation Method Manual (revised June 2011) with a slight modification [21].Based on the individual sample concentration, up to 12 libraries were pooled inequimolar amounts and processed according to the emPCR Amplification MethodManual (Lib-L, rev. June 2011). Deep sequencing was performed according to the

Please cite this article in press as: K. Uji et al., Significance of TP53 mutations dreceptor-positive breast cancer, Cancer Lett. (2013), http://dx.doi.org/10.1016/

Sequencing Method Manual (rev. June 2011). The average number of totalhigh-quality reads per run was 108,383, and the average number of coverage readswas 3108 per amplicon. (3) Analysis of deep sequencing: the processed and quality-filtered reads were analyzed using the GS Amplicon Variant Analyzer (AVA) soft-ware (version 2.7; Roche Diagnostics). The TP53 reference sequences were extractedfrom GenBank (Accession number: U94788). The percentage of mutant alleles wascalculated by dividing the number of mutant reads by the number of total readsusing the AVA software. The percentage of mutant alleles should be more than1% because the sensitivity of NGS, as reported by Moskalev et al. [17], is 0.5–1%.Those samples with homopolymer stretches and an unequal percentage of variantalleles between forward and reverse alignment were excluded.

2.4. RNA extraction and gene expression profiling

RNA was extracted from the tumor tissues obtained at surgery using the QiagenRNeasy� mini kit (Qiagen). The extracted RNA (1 lg; RIN value >6) was used for thegeneration of second-strand cDNA, and cRNA was amplified with the Oligo dT pri-mer, biotinylated, fragmented with the One-Cycle Target Labeling and control re-agents (Affymetrix, Santa Clara, CA, USA), and then hybridized overnight (17 h) toa U133 Plus 2.0 array according to the manufacturer’s protocol. The hybridizedDNA microarray was fluorescence stained with GeneChip� Fluidics Station 450and scanned with the GeneChip� Scanner 3000 (both from Affymetrix).

2.5. Immunohistochemical (IHC) examination

The expression of ER, the progesterone receptor (PR), Ki67, and p53 in tumortissues was immunohistochemically examined using a previously described meth-od [22,23]. ER, PR, and p53 were defined as positive if at least 10% of the tumor cellswere stained. Ki67 was defined as positive if at least 20% of the tumor cells werestained. The human epidermal growth factor receptor 2 (HER2) was identifiedimmunohistochemically (anti-human c-erbB-2 polyclonal antibody; Nichirei Bio-sciences, Tokyo, Japan) or through fluorescence in situ hybridization (FISH) usingPathVysion HER2 DNA probe kits (SRL Inc., Tokyo, Japan) for IHC +2 tumors. A tu-mor that exhibited +3 immunohistostaining or for which the FISH ratio was P2.0was considered HER2 positive. The histological grade was determined accordingto the Scarff–Bloom–Richardson grading system [24].

2.6. Microarray data processing

The RNA extracted from breast tumors was subjected to TP53 GES analysis usinga DNA microarray (Human Genome U133 plus 2.0 Array; Affymetrix) according to apreviously described method [19,20], and the breast tumors were classified as TP53GES-mutant and TP53 GES-wild-type tumors. The intrinsic subtypes (Luminal A,Luminal B, HER2-enriched, basal-like, and normal-like) were classified using thePAM50 method reported by Parker et al. [25]. For the PAM50 analysis, 26 ER-neg-ative tumors were added to the 115 ER-positive tumors because the PAM50 methodrequires a heterogeneous dataset that contains all subtypes.

2.7. Statistics

The association between the various clinicopathological parameters and theTP53 mutation status determined by SS or NGS were evaluated using the chi-squaretest or Fisher’s exact test. The differences in the frequency of TP53 mutant allelesbetween the TP53 mutations detected by SS and those detected by NGS were as-sessed with the Mann–Whitney test. The recurrence-free survival rates were calcu-lated with the Kaplan–Meier method and evaluated by the log-rank test (distantrecurrences and locoregional recurrences, except ipsilateral in-breast recurrences,were considered recurrence events). The univariate and multivariate analysis ofvarious parameters for the prediction of recurrences was conducted using theCox proportional hazards model. Regardless of the statistical test performed, differ-ences with P < 0.05 were considered statistically significant.

3. Results

3.1. Comparison of SS and NGS for the detection of TP53 mutations

The same DNA samples extracted from 115 breast tumors weresubjected to SS and NGS for the detection of TP53 mutations (Ta-ble 1). SS detected 20 TP53 mutations (SS-TP53 mutations), andNGS detected 27 TP53 mutations (NGS-TP53 mutations). Of the 27NGS-TP53 mutations, 20 were also detected by SS, and seven werenot. The frequency of mutant alleles was significantly (P < 0.001)higher in breast tumors with SS-TP53 mutations (n = 20, aver-age = 33.9%, range: 11.2–63.8%) compared with those with TP53mutations detected only by NGS (n = 7, average = 11.1%, range:1.6–24.1%). All of the missense mutations detected by NGS alone

etermined by next-generation ‘‘deep’’ sequencing in prognosis of estrogenj.canlet.2013.08.028


Table 1Cases with TP53 mutations detected by both SS and NGS (No. 1–20) and by NGS only (No. 21–27).

Case Sequence change Amino acid change Effect Mutant allele (%) No. of reads (mutant/total)

1 c.566C > T Ala ? Val Missense 63.8 1244/19492 c.31G > C Glu ? Gln Missense 63.0 476/7553 c.625_626delAG Frameshift 51.8 635/12274 c.787_880del94 Frameshift 37.8 648/17145 c.535C > A His ? Asn Missense 37.7 1786/47376 c.91 G > A Val ? Ile Missense 35.8 534/14937 c.145G > C Asp ? His Missense 34.3 176/5138 c.469G > T Val ? Phe Missense 34.2 877/25619 c.832 C > T Pro ? Ser Missense 32.6 592/1816

10 c.31G > C Glu ? Gln Missense 31.3 241/77011 c.600_608del9 In-frame deletion 30.7 1893/616612 c.997_1007del11 Frameshift 30.4 461/151813 c.711G > C Met ? Ile Missense 30.0 137/45714 c.743G > A Arg ? Gln Missense 29.2 393/134615 c.637C > T Arg ? Stop Nonsense 28.3 28/9916 c.448_460del13 Frameshift 25.6 982/383817 c.546_550del5 Frameshift 25.5 1577/617818 c.975_976insGG Frameshift 24.1 1548/642119 c.733G > A Gly ? Ser Missense 19.7 49/24920 c.1025 G > C Arg ? Pro Missense 11.2 105/94021 c.742C > T Arg ? Trp Missense 24.1 343/142622 c.920–1G > C Splicing 15.6 506/323523 c.731G > A Gly ? Asp Missense 13.5 155/114624 c.428T > A Val ? Glu Missense 9.3 186/199025 c.529_546del18 In-frame deletion 9.2 144/155826 c.743G > A Arg ? Gln Missense 4.6 164/360427 c.646G > A Val ? Met Missense 1.6 28/1785

SS, Sanger sequencing; NGS, next-generation sequencing.

(A) Sanger sequencing

1.5

1.0

0.5

(B) Next-generation sequencing

% of mutant allelec.646G>A

No mutant allele

▌A

▌C

▌G

▌T

▌-

Reference sequence

Fig. 1. Representative result of TP53 mutations detected by next-generation sequencing but not Sanger sequencing. (A) Sanger sequencing, (B) next-generation sequencing. Inthis case, the mutant allele (adenine) was not detected by Sanger sequencing but was detected by next-generation sequencing at a frequency of 1.6%.

K. Uji et al. / Cancer Letters xxx (2013) xxx–xxx 3

Please cite this article in press as: K. Uji et al., Significance of TP53 mutations determined by next-generation ‘‘deep’’ sequencing in prognosis of estrogenreceptor-positive breast cancer, Cancer Lett. (2013), http://dx.doi.org/10.1016/j.canlet.2013.08.028


Table 2Relationships between clinicopathological parameters and TP53 mutations detected by SS or NGS.

Characteristics Total Mutation status (SS) P-value Mutation status (NGS) P-value

TP53 wt TP53 mt TP53 wt TP53 mtn (%) n (%) n (%) n (%)

All cases 115 95 (82.6) 20 (17.4) 88 (76.5) 27 (23.5)

Age (years)<50 47 37 (78.7) 10 (21.3) 34 (72.3) 13 (27.7)P50 68 58 (85.3) 10 (14.7) 0.361a 54 (79.4) 14 (20.6) 0.379a

Tumor size (clinical)62 cm 65 60 (92.3) 5 (7.7) 56 (86.2) 9 (13.8)>2 cm 50 35 (70.0) 15 (30.0) 0.002a 32 (64.0) 18 (36.0) 0.005a

Nodal status (pathological)Negative 103 84 (81.6) 19 (18.4) 77 (74.8) 26 (25.2)Positive 12 11 (91.7) 1 (8.3) 0.689b 11 (91.7) 1 (8.3) 0.289b

Histological grade1 38 37 (97.4) 1 (2.6) 36 (94.7) 2 (5.3)2 62 50 (80.6) 12 (19.4) 47 (75.8) 15 (24.2)3 15 8 (53.3) 7 (46.7) 0.001a 5 (33.3) 10 (66.7) <0.001a

PRNegative 25 19 (76.0) 6 (24.0) 15 (60.0) 10 (40.0)Positive 90 76 (84.4) 14 (15.6) 0.373b 73 (81.1) 17 (18.9) 0.028a

HER2Negative 103 87 (84.5) 16 (15.5) 83 (80.6) 20 (19.4)Positive 12 8 (66.7) 4 (33.3) 0.218b 5 (41.7) 7 (58.3) 0.007b

Ki67Negative 92 78 (84.8) 14 (15.2) 73 (79.3) 19 (20.7)Positive 23 17 (73.9) 6 (26.1) 0.229b 15 (65.2) 8 (34.8) 0.153b

SS, Sanger sequencing; NGS, next-generation sequencing; wt, wild-type; mt, mutant; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2.a Chi-square test.b Fisher’s exact test.

Table 3Relationships of p53 immunohistochemistry, TP53 GES, and intrinsic subtypes with TP53 mutations detected by SS or NGS.

Characteristics Total Mutation status (SS) P-value Mutation status (NGS) P-value

TP53 wt TP53 mt TP53 wt TP53 mtn (%) n (%) n (%) n (%)

p53 ImmunohistochemistryNegative 91 84 (92.3) 7 (7.7) 82 (90.1) 9 (9.9)Positive 24 11 (45.8) 13 (54.2) <0.001b 6 (25.0) 18 (75.0) <0.001a

TP53 GESWild-type 74 69 (93.2) 5 (6.8) 66 (89.2) 8 (10.8)Mutant 41 26 (63.4) 15 (36.6) <0.001a 22 (53.7) 19 (46.3) <0.001a

Intrinsic subtypeLuminal A 46 44 (95.7) 2 (4.3) 44 (95.7) 2 (4.3)Luminal B 30 24 (80.0) 6 (20.0) 0.052a 24 (80.0) 6 (20.0) 0.052a

HER2-enriched 15 9 (60.0) 6 (40.0) 0.002a 7 (46.7) 8 (53.3) <0.001a

Basal-like 12 8 (66.7) 4 (33.3) 0.014a 5 (41.7) 7 (58.3) <0.001a

Normal-like 12 10 (83.3) 2 (16.7) 0.186a 8 (66.7) 4 (33.3) 0.014a

SS, Sanger sequencing; NGS, next-generation sequencing; wt, wild-type; mt, mutant; TP53 GES, TP53 gene expression signature.a Chi-square test.b Fisher’s exact test.


were previously reported (see TP53 Database, http://www-p53.iarc.fr/). A representative TP53 mutation that could be detectedby NGS but not by SS is shown in Fig. 1A and B.

3.2. Clinicopathological characteristics of breast tumors with TP53mutations

The relationships between the SS-TP53 mutations or the NGS-TP53 mutations and the various clinicopathological parametersare shown in Table 2. Breast tumors with SS-TP53 mutations aremore likely to be large (P = 0.002) and to have a high histologicalgrade (P = 0.001). Similarly, breast tumors with the NGS-TP53mutations were significantly associated with a large tumor size


(P = 0.005) and a high histological grade (P < 0.001). In addition,the NGS-TP53 mutations were also significantly associated withPR-negativity (P = 0.028) and HER2-positivity (P = 0.007).

The relationships between the SS-TP53 mutations or the NGS-TP53 mutations and the p53 immunohistochemistry, TP53 GES,and intrinsic subtypes are shown in Table 3. Breast tumors withthe SS-TP53 mutations are more likely to be p53-positive(P < 0.001) and TP53 GES-mutant tumors (P < 0.001). These associ-ations were also observed more significant with respect to theNGS-TP53 mutations, i.e., the percentages of tumors with TP53mutations among the p53-positive tumors were 54.2% for SS and75.0% for NGS and among the TP53 GES-mutant tumors were36.6% for SS and 46.3% for NGS. The frequency of breast tumors


http://www-p53.iarc.fr/

http://www-p53.iarc.fr/


Table 4Clinicopathological features, TP53 GES, and intrinsic subtypes of tumors with TP53 mutations detected only by NGS.

Case Mutation Histology Ta Nb HG PR HER2 Ki67 p53 IHC TP53 GESc Intrinsic subtyped

21 c.742C > T IDC 2 0 2 � � � + wt HER222 c.920–1G > C IDC 1 0 1 + � � � wt Normal23 c.731G > A IDC 1 0 3 + + + + mt HER224 c.428T > A Medullary ca. 1 0 2 + � � � mt Basal25 c.529–546del18 IDC 3 0 2 + � � + mt Basal26 c.743G > A IDC 1 0 3 � + � + wt Normal27 c.646G > A IDC 1 0 3 � + + + mt Basal

TP53 GES, TP53 gene expression signature; NGS, next-generation sequencing; HG, histological grade; PR, progesterone receptor; HER2, human epidermal growth factorreceptor 2; p53 IHC, p53 immunohistochemistry; IDC, invasive ductal carcinoma; medullary ca., medullary carcinoma.

a Clinical tumor category.b Pathological nodal status.c TP53 GES: wt, wild-type; mt, mutant.d Intrinsic subtypes: HER2, HER2-enriched; Normal, normal-like; Basal, basal-like.


with SS-TP53 mutations was significantly higher for the HER2-en-riched and basal-like subtypes compared with the luminal A sub-type. Breast tumors with the NGS-TP53 mutations also showedassociations with the HER2-enriched and basal-like subtypes witha greater statistical significance compared with those found for tu-mors with SS-TP53 mutations. In addition, the NGS-TP53 mutationsexhibited a significant association with the normal-like subtype.

Of the 115 breast tumors, seven possessed a TP53 mutation thatcould be detected by NGS but not by SS. The clinicopathologicalcharacteristics of these seven tumors, i.e., the tumor size, histolog-ical grade, PR, HER2, p53 expression, and TP53 GES, were similar tothose of tumors with the SS-TP53 mutations (Table 4). Interest-ingly, none of these seven tumors belonged to the luminal A orluminal B subtypes; two of them were classified as HER2-enriched,three were classified as basal-like, and two were classified as nor-mal-like. The analysis of the five tumors with the missense muta-tion detected only by NGS showed that almost all of the tumor cellsof four of these tumors were strongly positive for p53.

3.3. Prognostic significance of SS-TP53 mutations, NGS-TP53mutations, p53 immunohistostaining, and TP53 GES

Breast tumors with the SS-TP53 mutations showed significantly(P = 0.040) poorer prognosis compared with those without a muta-tion in the TP53 gene (Fig. 2A). However, no significant associationwas found between the NGS-TP53 mutations (Fig. 2B) or the p53immunohistochemistry (Fig. 2C) and prognosis. Patients withTP53 GES-mutant tumors exhibited significantly (P < 0.001) poorerprognosis compared with those with TP53 GES-wild-type tumors(Fig. 2D). The multivariate analysis showed that the prognostic sig-nificance of TP53 GES was independent of the other parameters(Table 5).

4. Discussion

The main purpose of this study was to use NGS for an analysis ofTP53 mutations in breast tumors because NGS is more sensitivethan SS and to obtain a better understanding of the clinicopatho-logical characteristics of breast tumors with a mutation in theTP53 gene. We were able to show that all of the SS-TP53 mutationscould also be detected by NGS and that NGS could identify sevenadditional mutations. This result clearly indicates that NGS is amore sensitive method for the detection of TP53 mutations. Inter-estingly, breast tumors with the TP53 mutations detected only byNGS were mostly (5 out of 7) classified as belonging to the nor-mal-like or basal-like subtypes. The normal-like subtype is knownto contain a very small amount of tumor cells largely contaminatedby normal breast tissue [26], and the basal-like subtype is oftenassociated with prominent lymphocyte infiltration. Thus, it is likely


that the inability of SS to detect this mutation is mostly attribut-able to contamination by normal breast tissue and/or lymphocytes.In fact, the frequency of mutant alleles in breast cancers with theTP53 mutations detected only by NGS (average: 11.1%) was signif-icantly (P < 0.001) lower than that obtained for breast cancers withthe SS-TP53 mutations (average: 33.9%). In addition, a mode in-depth analysis of the five breast cancers with the missense muta-tion detected only by NGS showed that almost all of the tumor cellsof four of these tumors were positive for p53, which indicates thatalmost all tumors cells carry this mutation. It thus appears unlikelythat the tumor tissue if composed of a very small fraction of tumorcells that carry the mutation and a large number of mutation-freetumor cells. Instead, it is more likely almost all of the tumor cellscarry the mutation but that the mutant alleles exhibit a low fre-quency due to contamination by a large amount of normal breasttissue and/or lymphocytes.

It has been well established that breast tumors with TP53 muta-tions are more likely to be characterized by a large tumor size andhigh histological grade and as p53-positive and TP53 GES-mutant[5,19,20]. This observation was confirmed in the breast tumorswith the SS-TP53 mutations. Interestingly, the statistical signifi-cance of these associations was found to be stronger for breast tu-mors with the NGS-TP53 mutations. Furthermore, the tendency ofbreast tumors with the SS-TP53 mutations to be associated withPR-negativity (P = 0.373) and HER2-positivity (P = 0.218) was sta-tistically significant for the NGS-TP53 mutations, i.e., P = 0.028 forPR-negativity and P = 0.007 for HER2-positivity. These observa-tions indicate that the NGS-TP53 mutations are in fact biologicallysignificant.

The SS-TP53 mutations are reportedly significantly associatedwith poor prognosis [5,29], and our finding that TP53 GES is astrong prognostic factor is consistent with that of previous reports[18,19]. TP53 GES appears to be more prognostic than the SS-TP53mutations because of a clearer statistically significant differentia-tion between low- and high-risk patients (P < 0.001 for TP53 GESand P = 0.040 for SS). There are many molecular mechanisms thatcompromise TP53 function. Thus, some deregulation of the TP53pathway can be captured by the TP53 GES but not through TP53mutation analysis, which indicates that the absence of the TP53mutation does not necessarily mean that the TP53 pathway is in-tact. This fact appears to explain, at least in past, why TP53-GESis more prognostic than the SS-analyzed TP53 mutations. We ex-pected that the more accurate detection of the TP53 mutationsby NGS would identify more mutations such that the NGS-TP53mutations would be more strongly associated with poor prognosisand would thus surpassing or at least equal the prognostic value ofthe TP53 GES. However, the NGS-TP53 mutations were unexpect-edly not a significant prognostic factor because none of the sevenpatients with breast tumors with the TP53 mutation detected only



(A) TP53 mutation status detected by SS (B) TP53 mutation status detected by NGS

(C) p53 Immunohistochemistry (D) TP53 gene expression signature

0 5 10 150

20

40

60

80

100

TP53 wild (n=95)TP53 mutant (n=20)

Log-rankP=0.040

Years after surgery

Rec

urre

nce-

Free

Sur

viva

l (%

)

0 5 10 150

20

40

60

80

100

TP53 wild (n=88)TP53 mutant (n=27)

Log-rankP=0.483

Years after surgery

Rec

urre

nce-

Free

Sur

viva

l (%

)

0 5 10 150

20

40

60

80

100

p53 IHC-negative (n=91)p53 IHC-positive (n=24)

Log-rankP=0.679

Years after surgery

Rec

urre

nce-

Free

Sur

viva

l (%

)

0 5 10 150

20

40

60

80

100

TP53 GES wild (n=74)TP53 GES mutant (n=41)

Log-rankP <0.001

Years after surgery

Rec

urre

nce-

Free

Sur

viva

l (%

)

Fig. 2. Prognostic significance of TP53 mutations determined by Sanger sequencing and next-generation sequencing, p53 immunohistochemistry, and TP53 gene expressionsignature. The recurrence-free survival rates of patients with ER-positive breast cancer treated with adjuvant hormonal therapy alone are shown in terms of the TP53mutation status determined by Sanger sequencing (A) or next-generation sequencing (B), the p53 immunohistochemistry (C), and the TP53 gene expression signature (GES)(D).

Table 5Univariate and multivariate analysis of various parameters associated with prognosis.

Univariate Multivariate

Hazard ratio 95% CI P-value Hazard ratio 95% CI P-value

Age (years) 0.622 0.311–1.245 0.180Tumor size (clinical) 2.156 1.064–4.368 0.033 2.091 0.976–4.478 0.058Nodal status (pathological) 1.725 0.709–4.199 0.230Histological grade 2.821 1.215–6.550 0.016 1.016 0.343–3.008 0.977PR 0.779 0.350–1.737 0.542HER2 3.675 1.583–8.530 0.002 2.530 0.884–7.242 0.084Ki67 1.838 0.849–3.979 0.122p53 IHC 0.829 0.341–2.016 0.680SS-detected TP53 mutation 2.201 1.016–4.769 0.045 0.891 0.370–2.146 0.796NGS-detected TP53 mutation 1.317 0.608–2.856 0.485Intrinsic subtype 3.274 1.374–7.958 0.009 1.438 0.495–4.180 0.505TP53 GES 4.382 2.111–9.097 <0.001 2.783 1.194–7.453 0.019

CI, confidence interval; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; p53 IHC, p53 immunohistochemistry; SS, Sanger sequencing; NGS, next-generation sequencing; TP53 GES, TP53 gene expression signature.Hazard ratio based on P50 versus <50; large tumor size (>2 cm) versus small tumor size (62 cm); lymph node metastasis: positive versus negative; histological grade IIIversus grade I + II; PR-positive versus PR-negative; HER2-positive versus HER2-negative; Ki67-positive versus Ki67-negative; p53 IHC-positive versus p53 IHC-negative; TP53mutant versus TP53 wild-type (detected by both SS and NGS); intrinsic subtype Luminal A versus other subtypes; TP53 GES mutant versus wild-type.


by NGS developed recurrence. The small number of patients ana-lyzed in the present study is the most plausible explanation forsuch a negative result. However, the possibility that breast tumorswith TP53 mutations detected only by NGS, which are character-ized by a low tumor cell content, may have a different biology interms of prognosis than that of breast tumors with the SS-TP53mutations, which are supposed to have a relatively high tumor cell


content, cannot be completely ruled out. The prognostic signifi-cance of the NGS-TP53 mutations thus needs to be clarified in a fu-ture study with a larger number of patients.

The prognostic significance of the immunohistochemicallydetermined p53 positivity remains controversial [30]. Of the 11breast tumors that were positive for p53 immunohistostainingbut negative for the SS-TP53 mutations, five were positive for the




NGS-TP53 mutations. However, six tumors remained positive forp53 immunohistostaining but negative for TP53 mutations. Be-cause NGS is presumed to be able to detect rare mutations witha high sensitivity, these six breast tumors are thought to overex-press wild-type p53 in the absence of a TP53 mutation. Thus, theimmunohistochemically determined p53 positivity is not an accu-rate method for the detection of TP53 mutations [31], which ex-plains, at least in part, the lack of association between p53positivity and poor prognosis found in the present study.

One limitation of the present study was its inability to comparethe tumor cell proportion in the tumor tissue with the detectabilityof TP53 mutations by SS or NGS. However, such a study appears tobe important to demonstrate the impact of contamination by non-cancerous cells on the detectability of TP53 mutations. In the pres-ent study, the tumor samples used for TP53 mutation analysis andhistological examination were obtained from different parts of thetumor; thus, their comparison is thought to be problematic be-cause a significant intra-tumoral heterogeneity of tumor cell pro-portion is often observed. Another limitation was that the loss ormaintenance of the TP53-wild-type allele could not be evaluatedin the TP53-mutant tumors by NGS analysis using the whole tumortissues due to the contamination of non-cancerous cells, whereasapproximately 50% of the TP53-mutant tumors exhibit the loss ofthe wild-type allele, as was determined in a previous study usingpolymorphic markers [27,28]. We hypothesize that the loss ormaintenance of the wild-type allele can be evaluated if the NGSanalysis is performed after the purification or enrichment of tumorcells from the whole tumor tissues.

In conclusion, we demonstrated that NGS is more sensitive tothe detection of TP53 mutations. However, the prognostic signifi-cance of these mutations could not be demonstrated. In contrast,the TP53 GES proved to be a powerful prognostic indicator forER-positive tumors. Our preliminary results highlight the impor-tance of using NGS in future studies of somatic mutations in breasttumors, which are often contaminated by a large amount of normaltissue, for a more accurate detection of these mutations.

Conflict of Interest

The authors of this study have no conflict of interest or anyfinancial disclosures to make.

Acknowledgments

This work was supported in part by Grants-in-Aid from theKnowledge Cluster Initiative of the Ministry of Education, Culture,Sports, Science, and Technology of Japan.

References

[1] L.J. Ko, C. Prives, P53: puzzle and paradigm, Genes Dev. 10 (1996) 1054–1072.[2] V.E. Velculescu, W.S. El-Deiry, Biological and clinical importance of the p53

tumor suppressor gene, Clin. Chem. 42 (1996) 858–868.[3] M.S. Greenblatt, W.P. Bennett, M. Hollstein, C.C. Harris, Mutations in the p53

tumor suppressor gene: clues to cancer etiology and molecular pathogenesis,Cancer Res. 54 (1994) 4855–4878.

[4] A.L. Borresen-Dale, TP53 and breast cancer, Hum. Mutat. 21 (2003) 292–300.[5] M. Olivier, A. Langerod, P. Carrieri, J. Bergh, S. Klaar, J. Eyfjord, C. Theillet, C.

Rodriguez, R. Lidereau, I. Bieche, J. Varley, Y. Bignon, N. Uhrhammer, R.Winqvist, A. Jukkola-Vuorinen, D. Niederacher, S. Kato, C. Ishioka, P. Hainaut,A.L. Borresen-Dale, The clinical value of somatic TP53 gene mutations in 1794patients with breast cancer, Clin. Cancer Res. 12 (2006) 1157–1167.

[6] P.D. Pharoah, N.E. Day, C. Caldas, Somatic mutations in the p53 gene andprognosis in breast cancer: a meta-analysis, Br. J. Cancer 80 (1999) 1968–1973.

[7] P. Bertheau, M. Espie, E. Turpin, J. Lehmann, L.F. Plassa, M. Varna, A. Janin, H. deThe, TP53 status and response to chemotherapy in breast cancer, Pathobiology75 (2008) 132–139.

[8] M. Varna, G. Bousquet, L.F. Plassa, P. Bertheau, A. Janin, TP53 status andresponse to treatment in breast cancers, J. Biomed. Biotechnol. (2011) 284–584.


[9] T. Aas, A.L. Borresen, S. Geisler, B. Smith-Sorensen, H. Johnsen, J.E. Varhaug, L.A.Akslen, P.E. Lonning, Specific P53 mutations are associated with de novoresistance to doxorubicin in breast cancer patients, Nat. Med. 2 (1996) 811–814.

[10] P. Bertheau, F. Plassa, M. Espie, E. Turpin, A. de Roquancourt, M. Marty, F.Lerebours, Y. Beuzard, A. Janin, H. de The, Effect of mutated TP53 on responseof advanced breast cancers to high-dose chemotherapy, Lancet 360 (2002)852–854.

[11] P. Bertheau, E. Turpin, D.S. Rickman, M. Espie, A. de Reynies, J.P. Feugeas, L.F.Plassa, H. Soliman, M. Varna, A. de Roquancourt, J. Lehmann-Che, Y. Beuzard,M. Marty, J.L. Misset, A. Janin, H. de The, Exquisite sensitivity of TP53 mutantand basal breast cancers to a dose-dense epirubicin–cyclophosphamideregimen, PLoS Med. 4 (2007) e90.

[12] E.M. Berns, J.A. Foekens, R. Vossen, M.P. Look, P. Devilee, S.C. Henzen-Logmans,I.L. van Staveren, W.L. van Putten, M. Inganas, M.E. Meijer-van Gelder, C.Cornelisse, C.J. Claassen, H. Portengen, B. Bakker, J.G. Klijn, Completesequencing of TP53 predicts poor response to systemic therapy of advancedbreast cancer, Cancer Res. 60 (2000) 2155–2162.

[13] H.S. Kim, C.K. Yom, H.J. Kim, J.W. Lee, J.H. Sohn, J.H. Kim, Y.L. Park, S.H. Ahn,Overexpression of p53 is correlated with poor outcome in premenopausalwomen with breast cancer treated with tamoxifen after chemotherapy, BreastCancer Res. Treat. 121 (2010) 777–788.

[14] S.J. Cleator, T.J. Powles, T. Dexter, L. Fulford, A. Mackay, I.E. Smith, H.Valgeirsson, A. Ashworth, M. Dowsett, The effect of the stromal componentof breast tumours on prediction of clinical outcome using gene expressionmicroarray analysis, Breast Cancer Res. 8 (2006) R32.

[15] A. Kohlmann, H.U. Klein, S. Weissmann, S. Bresolin, T. Chaplin, H. Cuppens, E.Haschke-Becher, B. Garicochea, V. Grossmann, B. Hanczaruk, K. Hebestreit, C.Gabriel, I. Iacobucci, J.H. Jansen, G. te Kronnie, L. van de Locht, G. Martinelli, K.McGowan, M.R. Schweiger, B. Timmermann, P. Vandenberghe, B.D. Young, M.Dugas, T. Haferlach, The Interlaboratory RObustness of Next-generationsequencing (IRON) study: a deep sequencing investigation of TET2, CBL andKRAS mutations by an international consortium involving 10 laboratories,Leukemia 25 (2011) 1840–1848.

[16] F. Buttitta, L. Felicioni, M. Del Grammastro, G. Filice, A. Di Lorito, S. Malatesta,P. Viola, I. Centi, T. D’Antuono, R. Zappacosta, S. Rosini, F. Cuccurullo, A.Marchetti, Effective assessment of EGFR mutation status in bronchoalveolarlavage and pleural fluids by next-generation sequencing, Clin. Cancer Res. 19(2013) 691–698.

[17] E.A. Moskalev, R. Stohr, R. Rieker, S. Hebele, F. Fuchs, H. Sirbu, S.E. Mastitsky, C.Boltze, H. Konig, A. Agaimy, A. Hartmann, F. Haller, Increased detection rates ofEGFR and KRAS mutations in NSCLC specimens with low tumour cell contentby 454 deep sequencing, Virchows Arch. 462 (2013) 409–419.

[18] L.D. Miller, J. Smeds, J. George, V.B. Vega, L. Vergara, A. Ploner, Y. Pawitan, P.Hall, S. Klaar, E.T. Liu, J. Bergh, An expression signature for p53 status in humanbreast cancer predicts mutation status, transcriptional effects, and patientsurvival, Proc. Natl. Acad. Sci. USA 102 (2005) 13550–13555.

[19] S. Takahashi, T. Moriya, T. Ishida, H. Shibata, H. Sasano, N. Ohuchi, C. Ishioka,Prediction of breast cancer prognosis by gene expression profile of TP53 status,Cancer Sci. 99 (2008) 324–332.

[20] K. Oshima, Y. Naoi, K. Kishi, Y. Nakamura, T. Iwamoto, K. Shimazu, T.Nakayama, S.J. Kim, Y. Baba, Y. Tamaki, S. Noguchi, Gene expressionsignature of TP53 but not its mutation status predicts response to sequentialpaclitaxel and 5-FU/epirubicin/cyclophosphamide in human breast cancer,Cancer Lett. 307 (2011) 149–157.

[21] Q. Jiang, T. Turner, M.X. Sosa, A. Rakha, S. Arnold, A. Chakravarti, Rapid andefficient human mutation detection using a bench-top next-generation DNAsequencer, Hum. Mutat. 33 (2012) 281–289.

[22] K. Morimoto, S.J. Kim, T. Tanei, K. Shimazu, Y. Tanji, T. Taguchi, Y. Tamaki, N.Terada, S. Noguchi, Stem cell marker aldehyde dehydrogenase 1-positivebreast cancers are characterized by negative estrogen receptor, positivehuman epidermal growth factor receptor type 2, and high Ki67 expression,Cancer Sci. 100 (2009) 1062–1068.

[23] A. Shimomura, Y. Miyoshi, T. Taguchi, Y. Tamaki, S. Noguchi, Association of lossof BRCA1 expression with centrosome aberration in human breast cancer, J.Cancer Res. Clin. Oncol. 135 (2009) 421–430.

[24] H.J. Bloom, W.W. Richardson, Histological grading and prognosis in breastcancer; a study of 1409 cases of which 359 have been followed for 15 years, Br.J. Cancer 11 (1957) 359–377.

[25] J.S. Parker, M. Mullins, M.C. Cheang, S. Leung, D. Voduc, T. Vickery, S. Davies, C.Fauron, X. He, Z. Hu, J.F. Quackenbush, I.J. Stijleman, J. Palazzo, J.S. Marron, A.B.Nobel, E. Mardis, T.O. Nielsen, M.J. Ellis, C.M. Perou, P.S. Bernard, Supervisedrisk predictor of breast cancer based on intrinsic subtypes, J. Clin. Oncol. 27(2009) 1160–1167.

[26] F. Elloumi, Z. Hu, Y. Li, J.S. Parker, M.L. Gulley, K.D. Amos, M.A. Troester,Systematic bias in genomic classification due to contaminating non-neoplastictissue in breast tumor samples, BMC Med. Genomics 4 (2011) 54.

[27] A.M. Thompson, T.J. Anderson, A. Condie, J. Prosser, U. Chetty, D.C. Carter, H.J.Evans, C.M. Steel, P53 allele losses, mutations and expression in breast cancerand their relationship to clinico-pathological parameters, Int. J. Cancer 50(1992) 528–532.

[28] G. Deng, L.C. Chen, D.R. Schott, A. Thor, V. Bhargava, B.M. Ljung, K. Chew, H.S.Smith, Loss of heterozygosity and p53 gene mutations in breast cancer, CancerRes. 54 (1994) 499–505.

[29] J. Bergh, T. Norberg, S. Sjogren, A. Lindgren, L. Holmberg, Complete sequencingof the p53 gene provides prognostic information in breast cancer patients,


http://refhub.elsevier.com/S0304-3835(13)00603-4/h0005













































































































particularly in relation to adjuvant systemic therapy and radiotherapy, Nat.Med. 1 (1995) 1029–1034.

[30] L. Harris, H. Fritsche, R. Mennel, L. Norton, P. Ravdin, S. Taube, M.R. Somerfield,D.F. Hayes, R.C. Bast Jr., American Society of Clinical Oncology 2007 update ofrecommendations for the use of tumor markers in breast cancer, J. Clin. Oncol.25 (2007) 5287–5312.


[31] S. Sjogren, M. Inganas, T. Norberg, A. Lindgren, H. Nordgren, L. Holmberg, J.Bergh, The p53 gene in breast cancer: prognostic value of complementary DNAsequencing versus immunohistochemistry, J. Natl. Cancer Inst. 88 (1996) 173–182.













significance of tp53 mutations determined by next-generation “deep” sequencing in prognosis of...

Documents