Etiologic value of p53 mutation spectra and differences with histology in lung cancers

Abstract

A total of 297 resected Japanese non‐small cell lung cancers (74 squamous cell carcinomas and 223 adenocarcinomas) were analyzed to evaluate the validity of the p53 mutation spectrum as a fingerprint for mutagenic substances as etiological factors. Frequencies of G→T transversions in smokers were significantly higher than in non‐smokers (P = 0.003) and the average incidence of G→T at hot spot codons of adduct formation was higher than that in other codons in smokers and in the hot spots in non‐smokers. Further, the mutation showed a marked strand bias. G→A transitions at CpG sites (CpG→CpA) were equally distributed in smokers and non‐smokers, and on both strands. A→G transitions did not show any variation with smoking status in terms of frequency, but exhibited a marked strand bias. Taken together, the G→T may be a fingerprint of direct mutagenic action of tobacco‐related compounds, the A→G being a new marker for other environmental chemicals, while the CpG→CpA may be attributable to endogenous spontaneous mutation, for active in lung carcinogenesis. (Cancer Sci 2008; 99: 287–295)

The p53 tumor suppressor gene plays an important role in prevention of carcinoma development through its apoptotic and cell cycle checkpoint functions.^{( 1} , ^{2 )} Mutations of the gene have been widely recognized in many kinds of human tumors, and mutation patterns are considered to offer ‘fingerprints’ for mutagenic substances. The high frequency of G→T transversions in lung cancers has been attributed to the direct mutagenic action of tobacco smoke components, in particular polycyclic aromatic hydrocarbons (PAHs).^{( 3} , ⁴ , ⁵ , ⁶ , ⁷ , ^{8 )} In contrast, most G→A transitions at CpG sites (CpG→CpA) are ascribed to endogenous mechanisms, because they are presumed to arise due to spontaneous hydrolytic deamination of cytosine at methylated CpG sites.^{( 3} , ⁹ , ¹⁰ , ^{11 )}

However, other notions on the genesis of mutation patterns have recently been presented. Rodin and Rodin have questioned the direct mutagenic action of PAH‐like compounds and have suggested that other factors, such as selection of pre‐existing endogenous mutations by physiological stress aggravated by smoking^{( 12 )} can better explain the excess of G→T transversions in lung tumors. Two different ideas also exist for the causes of G→T transversions – direct mutagenic action or selection of pre‐existing endogenous mutations – from both in vivo and in vitro studies.^{( 13} , ^{14 )} Relationships between p53 mutation patterns and smoking status are critical for judgment of etiological influences.

Many authors have analyzed the International Agency for Research on Cancer (IARC) database for lung cancers to elucidate relationships, but have reached different conclusions. One reason may partly depend on the fact that the database, as described by Hainaut et al. is exclusively a repository for mutations described in peer‐reviewed articles,^{( 8 )} and some information on smoking status, occupational exposure to known carcinogenic agents, tumor histology, sex, and ethnicity, which confound the relationship between p53 mutation spectra and smoking, is uncertain. As to CpG→CpA transitions, the DNA sequence context may participate in the generation of a mutation pattern, lesions being found to occur predominantly in a 5′‐CGT‐3′ sequence context with activation of benzo[a]pyrene (BaP) in vitro.^{( 15 )} Adducts such as those formed with alkylating agents are another likely cause of G→A transitions.^{( 16} , ¹⁷ , ^{18 )} However, to our knowledge there has been no detailed study on the genesis of CpG→CpA transitions using human lung cancers.

It is important to judge whether the p53 mutation patterns in lung cancers (e.g. those involving G→T transversions and CpG→CpA transitions) are induced by direct mutagenic action of inhaled exogenous carcinogens, especially tobacco smoke compounds, or endogenous processes, respectively, for identification of carcinogenic agents and thus, clues to prevention methods.

We have collected a large series of cases of Japanese non‐small cell lung cancers (NSCLCs) with accurate smoking status, undergoing surgery at one hospital that were classified histologically by the same pathologists. The cases were examined for p53 mutation spectra to elucidate relationships, especially for G→T transversions and CpG→CpA transitions, with smoking status to re‐evaluate the validity of fingerprints for mutagenic substances with Japanese NSCLCs. Further, to clarify etiological differences between squamous cell carcinomas (SQCCs) and adenocarcinomas (ADCs) of the lung, a comparison of their mutation spectra was carried out.

Materials and Methods

Tumor samples, clinicopathological data, and smoking history.  We examined a large series of 297 Japanese NSCLCs (223 ADCs and 74 SQCCs) that had been consecutively resected from 1989 to 1995 at Cancer Institute Hospital, Tokyo, Japan. All patients analyzed had undergone a potentially curative resection with lobectomy or pneumonectomy, combined with pulmonary hilar and mediastinal systematic lymph node dissection. Out of 223 ADC cases, three received preoperative chemotherapy, and 67 had postoperative therapy (adjuvant chemotherapy for 49, local radiotherapy for 16, and both for two). With the SQCC patients, none had received chemotherapy and radiotherapy before surgery, but 25 underwent postoperative adjuvant chemotherapy. The study population was aged 26–84, with a mean of 62 years, and a mean age of the SQCC cases (66) was higher than that for ADCs (61) (Table 1). The patients were 193 males and 104 females in total, but since the number of females with SQCCs was very small (only five) the figure of 99 ADCs was almost equal to the 124 ADCs in males. Histological diagnosis was carried out on the basis of the 1999 World Health Organization classification of lung tumors by two of the authors (E.T. and Y.I.). The pathological stages (p‐stages) were determined using the Union Internationale Contre le Cancer (UICC) tumor node metastasis staging system, sixth edition. Most cases were in p‐stage I or III, and the same was observed for adenocarcinomas.

Table 1.

p53 mutations and clinicopathological parameters

Smoking histories (number of cigarettes per day, starting age, and duration of smoking) were obtained from preoperative personal interviews, with division into never‐smokers and smokers, the latter including both former and current smokers. To evaluate the amount of cigarette consumption, a smoking index (SI) was used: cigarette consumption per day multiplied by smoking years. Referring to the index, smokers were divided into two groups, heavy smokers with indices ≥400, and light ones <400. The percentage of smokers for all cases was 64% (191/297) (87% for males and 22% for females). When looked at by histology, the percentages showed marked differences; 89% (66/74) for SQCCs, compared to 56% (125/223) for ADCs (P < 0.001). Most smokers were heavy smokers, and the rate was higher for SQCCs (95%, 63/66) than for ADCs (81%, 101/125) (P < 0.01).

DNA preparation.  Fresh tumor samples were obtained from all patients, quickly frozen in liquid nitrogen, and stored at –80°C until DNA extraction and analysis. Genomic DNAs for SQCC samples were prepared as previously described.^{( 19 )} For ADCs, the DNAs for part of the tumor samples (the first 124) were prepared as previously described.^{( 20 )} For the latter part (115) and for 78 samples from the first part that did not show p53 mutations, DNAs were extracted from microdissected tissues. Frozen specimens were cut serially at 25 µm and sections were placed in 99% ethanol. Microdissection was carried out manually under direct observation with a stereoscope for two to four sections stained with Hematoxylin using 18G or 22G needles and the collected tumor cells in 99% ethanol were pelleted by centrifugation at high speed (13 000 g ) for 5 min. DNAs were extracted using a DNA extraction kit (Puregene Kit, Gentra Systems, MN, USA) according to the manufacturer's instructions.

PCR‐SSCP, and Sequencing.  Exons 4–8 and 10 of the p53 gene were analyzed in all cases. The polymerase chain reaction (PCR)‐SSCP method and sequencing were carried out for SQCCs and the first half of ADCs with the primers and PCR conditions described previously.^{( 20} , ^{21 )} For the second half of the ADCs and those that did not show p53 mutation in the first half, PCR amplification reactions and direct sequencing methods were used.

Sequences of oligonucleotides for PCR were as follows: exon 4 of p53, the sense primer, 5′‐ACC TGG TCC TCT GAC TGC TCT TTT CA and the antisense primer, 5′‐CCA GGC ATT GAA GTC TCA TGG AAG C; exon 5–8, 5′‐CTG TTC ACT TGT GCC CTG ACT TTC AAC and 5′‐TCT GAG GCA TAA CTG CAC CCT TGG TCT; exon 10, 5′‐TAT ACT TAC TTC TCC CCC TCC TCT and 5′‐ATG AGA ATG GAA TCC TAT GGC TTT. PCR amplification of exons 5–8 was carried out together due to their close to proximity. PCR reaction mixtures contained 50 ng genomic DNA, 10 pM of each pair of primers, 10 mM deoxynucleotide triphosphates (dATP, dTTP, dGTP and dCTP), 1.0 U Taq DNA polymerase (Platinum Taq DNA polymerase High Fidelity, Invitrogen), 600 mM Tris‐SO₄ (pH 8.9) buffer, 2 mM magnesium sulfate. After initial denaturation (for 2 min at 94°C), 35–40 cycles of PCR were carried out as follows: 94°C for 30 s, adequate annealing temperature 60°C for 30 s and 68°C for 30 s with a T‐personal thermal cycler system (Biometra).

After PCR product purification by centrifugal filtration (Montage MILLIPORE), direct sequencing was carried out for exons 4, 5–8 and 10 of the p53 gene with a CEQ 2000 DNA analyzer (Beckman Coulter Inc.) using CEQ 2000 Dye terminator cycle sequencing and a quick start kit (Beckman Coulter Inc.).

Statistical analysis.  To assess any correlations between the p53 mutation status and clinicopathological data, the χ² test, Fisher's exact probably test, Student's t‐test, and Mann–Whitney's U‐test were used, with significance concluded at P < 0.05.

Results

p53 mutations.  Of the 297 NSCLCs, 142 (48%) had p53 mutations; 46 of the SQCCs and 96 of the ADCs (Table 1). Seven cases (DNA nos. 25, 31 and 43 in SQCCs, and DNA nos. 17, 244, 350 and 360 in ADCs) had two mutations each, making 149 (50%) in total (Table 2). Fourteen mutations (9%) were located in exon 4, 36 (24%) in exon 5, 26 (17%) in exon 6, 33 (22%) in exon 7, 28 (19%) in exon 8, five (3%) in exon 10, and seven (5%) in splicing junctions of exons. The majority (134, 90%) was located in the sequence‐specific DNA binding region (codons 100–293). There were 11 codons, 157, 158, 175, 176, 196, 220, 245, 248, 249, 273 and 282, where the number of mutations were more than three (Fig. 1a,b; Table 2). All six known mutation hotspots for lung cancer were included; 11 at codon 273, nine at 245, six at 248, five at 158, and three each at 157 and 249, and this total of 37 accounted for 25% for all mutations.

Table 2.

p53 mutations in lung adenocarcinomas and squamous cell carcinomas

Sequential no.	DNA no.	Age (years)	Sex	Smoking index †	Tumor		Exon	Mutation					Amino acid	Mutation pattern
								Codon		Base change by strand		Next letter of the mutation to 3′ on the transcribed strand
					Histology	pStage		No.	Letters	Non‐transcribed	Transcribed
1	198	55	M	700	Ad	IA	4	46	Repeat of 16 bp			−	Frameshift	Insertion
2	393	54	F	0	Ad	IIIB	4	98	Deletion of 1bp			−	Frameshift	Deletion
3	360	64	M	1760	Ad	IIB	4	104	CAG→CAT	G→T		−	Gln→His	TV
4	360	64	M	1760	Ad	IIB	4	105	GGC→TGC	G→T		−	Gly→Cys	TV
5	269	66	M	920	Ad	IIIB	4	110	Deletion of 1bp	CGT→GT		−	Frameshift	Deletion
6	138	44	M	0	Ad	IIIA	4	120	AAG→AGG	A→G		−	Lys→Arg	TS
7	316	60	M	0	Ad	IA	4	124	Insertion of 1bp			−	Frameshift	Insertion
8	136	50	M	1600	Ad	IB	4	125	ACG→ACT	G→T		−	Thr→Thr	TV
9	363	63	M	540	Ad	IIIB	4	125	ACG→ACA	G→A (CpG)		−	Thr→Thr	TS
10	111	54	M	540	Ad	IB	4	3′ junction	Deletion of 1bp	CGgt→CG_t		−	Splicing	Deletion
11	368	73	F	0	Ad	IIIB	4	41–53	Deletion of 38bp			−	Frameshift	Deletion
12	17	69	M	846	Ad	IA	4	113–119	Deletion of 19bp			−	Frameshift	Deletion
13	90	73	M	1040	Ad	IB	4	124	Deletion of 25bp			−	Frameshift	Deletion
14	203	67	F	0	Ad	IIIB	5	132	AAG→AGG	A→G		−	Lys→Arg	TS
15	307	66	M	740	Ad	IIIA	5	132	AAG→GAG	A→G		−	Lys→Gln	TS
16	105	61	M	1600	Ad	IB	5	135	TGC→TTC	G→T		−	Cys→Phe	TV
17	255	66	F	0	Ad	IIIB	5	136	CAA→TAA	C→T	G→A (non‐CpG)	−	Gln→Stop	TS
18	11	72	F	0	Ad	IIIB	5	138	GCC→GTC	C→T	G→A (non‐CpG)	−	Ala→Val	TS
19	19	57	F	0	Ad	IIIA	5	138	GCC→CCC	G→C		−	Ala→Pro	TV
20	208	71	M	990	Ad	IIIA	5	157	GTC→TTC	G→T		−	Val→Phe	TV
21	290	69	M	2820	Ad	IIIB	5	157	GTC→GGC	T→G	A→C	−	Val→Gly	TV
22	22	72	M	800	Ad	IIIA	5	158	CGC→CAC	G→A (CpG)		−	Arg→His	TS
23	96	60	M	800	Ad	IIA	5	158	CGC→CAC	G→A (CpG)		−	Arg→His	TS
24	173	54	M	680	Ad	IA	5	158	CGC→CTC	G→T		−	Arg→Leu	TV
25	197	47	M	1620	Ad	IIIA	5	158	CGC→CCC	G→C		−	Arg→Pro	TV
26	389	57	F	0	Ad	IIA	5	158	CGC→CAC	G→A (CpG)		−	Arg→His	TS
27	79	56	M	720	Ad	IIIB	5	159	Deletion of 2bp	GCC→C		−	Frameshift	Deletion
28	364	70	M	800	Ad	IB	5	159	GCC→GTC	C→T	G→A (non‐CpG)	−	Ala→Val	TS
29	315	81	M	0	Ad	IIIB	5	164	AAG→TAG	A→T		−	Lys→Stop	TV
30	293	60	M	1200	Ad	5	167	Deletion of 2bp	CAG→ G	−			Frameshift	Deletion
31	238	71	M	765	Ad	IIIB	5	171	GAG→TAG	G→T		−	Glu→Stop	TV
32	278	51	F	0	Ad	IV	5	172	Deletion of 2bp	GTT→T		−	Frameshift	Deletion
33	103	74	M	860	Ad	IA	5	175	CGC→CAC	G→A (CpG)		−	Arg→His	TS
34	134	26	F	0	Ad	IIIA	5	176	TGC→TTC	G→T		−	Cys→Phe	TV
35	399	51	F	420	Ad	IB	5	176	TGC→AGC	T→A	A→T	−	Cys→Ser	TV
36	142	70	M	561	Ad	IB	5	181	CGC→CCC	G→C		−	Arg→Pro	TV
37	352	64	F	0	Ad	IIIB	5	169–170	Insertion of 6bp			−	Inframe	Insertion
38	191	50	M	1160	Ad	IA	5	179–185	Deletion of 18bp			−	Inframe	Deletion
39	398	59	F	0	Ad	IIA	6	188	Duplication			−	Frameshift	Duplication
40	302	51	M	620	Ad	IB	6	189	GCC→GTC	C→T	G→A (non‐CpG)	−	Ala→Val	TS
41	160	50	M	2310	Ad	IIIB	6	190	Deletion of 1bp	CCT→CT	−		Frameshift	Deletion
42	380	38	F	0	Ad	IIA	6	194	CTT→CCT	T→C	A→G	−	Leu→Pro	TS
43	391	81	F	0	Ad	IV	6	196	CGA→TGA	C→T	G→A (CpG)	−	Arg→Stop	TS
44	97	54	M	960	Ad	IIIA	6	198	GAA→TAA	G→T		−	Glu→Stop	TV
45	361	69	M	920	Ad	IIIA	6	205	TAT→TGT	A→G		−	Tyr→Cys	TS
46	122	74	F	240	Ad	IIIA	6	209	AGA→TGA	A→T		−	Arg→Stop	TV
47	205	49	F	0	Ad	IB	6	213	CGA→TGA	C→T	G→A (CpG)	A	Arg→Stop	TS
48	329	53	M	100	Ad	IIA	6	215	AGT→ATT	G→T		−	Ser→lle	TV
49	357	68	F	0	Ad	IIIA	6	218	Deletion of 1bp	−			Frameshift	Deletion
50	350	63	F	0	Ad	IIIA	6	219	CCC→TCC	C→T	G→A (non‐CpG)	−	Pro→Ser	TS
51	77	41	F	0	Ad	IA	6	220	TAT→TGT	A→G	−		Tyr→Cys	TS
52	186	74	M	1060	Ad	IB	6	3′ junction	AGgt→AGat	G→A (non‐CpG)		−	Splicing	TS
53	89	41	M	750	Ad	IB	6	5′ junction	agG→atG	G→T		−	Splicing	TV
54	23	58	M	25	Ad	IA	7	234	TAC→TGC	A→G		−	Tyr→Cys	TS
55	38	77	F	0	Ad	IIIB	7	237	ATG→ATT	G→T		−	Met→lle	TV
56	86	56	M	900	Ad	IIIA	7	238	TGT→AGT	T→A	A→T	−	Cys→Ser	TV
57	157	49	M	20	Ad	IIA	7	238	TGT→AGT	T→A	A→T	−	Cys→Ser	TV
58	344	46	F	230	Ad	IIIB	7	239	AAC→GAC	A→G		−	Asn→Asp	TS
59	80	65	F	0	Ad	IA	7	241	Deletion of 1bp	TCC→TC		−	Frameshift	Deletion
60	101	68	F	75	Ad	IIIA	7	242	TGC→TAC	G→A (non‐CpG)		−	Cys→Tyr	TS
61	294	58	M	1440	Ad	IA	7	244	GGC→TGC	G→T		−	Gly→Cys	TV
62	28	47	M	650	Ad	IIIA	7	245	GGC→TGC	G→T		−	Gly→Cys	TV
63	66	51	F	0	Ad	IIIB	7	245	GGC→AGC	G→A (CpG)		−	Gly→Ser	TS
64	235	61	F	300	Ad	IIIB	7	245	GGC→CGC	G→C		−	Gly→Arg	TV
65	313	61	M	2420	Ad	IB	7	245	GGC→TGC	G→T		−	Gly→Cys	TV
66	381	61	M	1600	Ad	IIIB	7	245	GGC→GTC	G→T		−	Gly→Val	TV
67	33	37	F	30	Ad	IA	7	248	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
68	139	70	F	0	Ad	IA	7	248	CGG→CAG	G→A (CpG)		−	Arg→Gln	TS
69	297	59	M	570	Ad	IA	7	248	CGG→CAG	G→A (CpG)		−	Arg→Gln	TS
70	327	64	M	1020	Ad	IB	7	249	AGG→ATG	G→T		−	Arg→Met	TV
71	350	63	F	0	Ad	IIIA	7	258	GAA→GAC	A→C		−	Glu→Asp	TV
72	3	73	M	700	Ad	IIIB	7	259	GAC→AAC	G→A (non‐CpG)		−	Asp→Asn	TS
73	244	62	M	1000	Ad	IIIA	7	259	GAC→TAC	G→T		−	Asp→Tyr	TV
74	282	66	M	1290	Ad	IV	7	253‐254	Deletion of 3bp			−	Inframe	Deletion
75	49	49	F	0	Ad	IIIA	8	273	CGT→CAT	G→A (CpG)		−	Arg→His	TS
76	50	70	M	540	Ad	IIIA	8	273	CGT→CAT	G→A (CpG)		−	Arg→His	TS
77	69	58	M	1600	Ad	IIIA	8	273	CGT→TGT	C→T	G→A (CpG)	C	Arg→Cys	TS
78	100	48	M	900	Ad	IA	8	273	CGT→CTT	G→T		−	Arg→Leu	TV
79	152	68	F	0	Ad	IA	8	273	CGT→CTT	G→T		−	Arg→Leu	TV
80	174	72	M	510	Ad	IA	8	273	CGT→CAT	G→A (CpG)		−	Arg→His	TS
81	182	56	M	750	Ad	IA	8	273	CGT→TGT	C→T	G→A (CpG)	C	Arg→Cys	TS
82	215	63	M	0	Ad	IA	8	273	CGT→CTT	G→T		−	Arg→Leu	TV
83	259	56	M	435	Ad	IIIB	8	273	CGT→TGT	C→T	G→A (CpG)	C	Arg→Cys	TS
84	362	53	M	660	Ad	IIIB	8	273	CGT→TGT	C→T	G→A (CpG)	C	Arg→Cys	TS
85	34	59	M	480	Ad	IA	8	274	GTT→TTT	G→T		−	Val→Phe	TV
86	156	65	M	0	Ad	IA	8	275	TGT→TAT	G→A (non‐CpG)		−	Cys→Tyr	TS
87	244	62	M	1000	Ad	IIIA	8	281	GAC→TAC	G→T		−	Asp→Tyr	TV
88	155	63	F	0	Ad	IIIA	8	282	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
89	382	54	F	0	Ad	IB	8	282	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
90	400	61	M	820	Ad	IB	8	286	GAA→GTA	A→T		−	Glu→Val	TV
91	347	80	F	240	Ad	IIIA	8	287	GAG→TAG	G→T		−	Glu→Stop	TV
92	331	63	M	840	Ad	IA	8	298	GAG→TAG	G→T		−	Glu→Stop	TV
93	154	72	M	2520	Ad	IIIB	8	274	Deletion of 2bp	GTT→T		−	Frameshift	Deletion
94	17	69	M	846	Ad	IA	8	301	Deletion of 1bp	CCA→C A		−	Frameshift	Deletion
95	284	75	M	2120	Ad	IIIA	8	–262	Deletion of 17bp			−	Splicing	Deletion
96	15	64	M	660	Ad	IA	8	3′ junction	AGgt→AGtt	G→T		−	Splicing	TV
97	185	67	M	612	Ad	IIB	8	305–306	Repeat of 23bp			−	Frameshift	Insertion
98	83	65	F	0	Ad	IA	10	335	CGT→CAT	G→A (CpG)		−	Arg→His	TS
99	388	79	F	0	Ad	IIIB	10	342	CGA→TGA	C→T	G→A (CpG)	G	Arg→Stop	TS
100	148	51	F	0	Ad	IIIA	10	341	Deletion of 1bp	TTC→T C		−	Frameshift	Deletion
101	25	71	M	1020	Sq	IIIB	4	88	GCC→ACC	G→A (non CpG)		−	Ala→Thr	TS
102	24	71	M	1020	Sq	IIIB	4	103	TAC→TAG	C→G	G→C	−	Tyr→stop	TV
103	36	69	M	1020	Sq	IIB	5	130	CTC→GTC	C→G	G→C	−	Leu→Val	TV
104	16	76	M	1250	Sq	IIB	5	144	CAG→CCG	A→C	−	−	Gln→Pro	TV
105	43	76	M	780	Sq	IIIB	5	152	Deletion of 2bp	CCG→G	−	−	Frameshift	Deletion
106	33	75	M	1060	Sq	IB	5	154	GGC→GTC	G→T	−	−	Gly→Val	TV
107	44	79	M	0	Sq		5	157	GTC→TTC	G→T	−	−	Val→Phe	TV
108	30	71	M	2550	Sq	IB	5	163	TAC→TGC	A→G	−	−	Try→Cys	TS
109	21	69	M	2040	Sq	IA	5	166	TCA→TAA	C→A	G→T	−	Ser→stop	TV
110	9	59	M	1200	Sq	IIIA	5	175	CGC→CAC	G→A (CpG)	−	−	Arg→His	TS
111	28	60	M	1600	Sq	IIIB	5	175	CGC→CAC	G→A (CpG)	−	−	Arg→His	TS
112	27	67	M	470	Sq	IA	5	176	TGC→TAC	G→A (non CpG)	−	−	Cys→Tyr	TS
113	7	70	M	1250	Sq	IIIB	5	149‐175	Deletion of 79bp		−	−	Frameshift	Deletion
114	37	71	M	820	Sq	IB	5	5'junction	agTA→tgTA	a→t	−	−	splicing	TV
115	14	63	M	1505	Sq	IIIA	6	190	Deletion of 1bp	CCT→CT	−	−	Frameshift	Deletion
116	45	66	M	0	Sq	6	190	CCT→CTT	C→T	G→A (nonCpG)		−	Pro→Leu	TS
117	26	81	M	1100	Sq	IB	6	193	CAT→TAT	C→T	G→A (nonCpG)	−	His→Tyr	TS
118	19	70	M	1590	Sq	IIIA	6	195	ATC→ACC	T→C	A→G	−	lle→Thr	TS
119	42	69	M	960	Sq	IIIA	6	195	ATC→ACC	T→C	A→G	−	Ile→Thr	TS
120	15	49	M	300	Sq	IIIA	6	196	CGA→CCA	G→C	−	−	Arg→Pro	TV
121	35	65	M	1260	Sq	IIB	6	196	CGA→TGA	C→T	G→A (CpG)	G	Arg→stop	TS
122	2	71	M	0	Sq	IA	6	220	TAT→TGT	A→G	−	−	Try→Cys	TS
123	18	82	M	1170	Sq	IIIB	6	220	TAT→TGT	A→G	−	−	Try→Cys	TS
124	31	62	M	1320	Sq	IIB	6	220	TAT→TTT	A→T	−	−	Try→Cys	TV
125	38	61	M	1200	Sq	IB	6	220	TAT→TGT	A→G	−	−	Tyr→Cys	TS
126	41	66	M	966	Sq	IIIA	6	220	TAT→TGT	A→G	−	−	Tyr→Cys	TS
127	31	62	M	1320	Sq	IIB	6	221	Deletion of 1bp	GAG→AG	−	−	Frameshift	Deletion
128	40	69	M	490	Sq	IIIB	7	234	TAC→TGC	A→G	−	−	stop→Tro	TS
129	6	64	M	0	Sq	IIIA	7	236	TAC→TGC	A→G	−	−	Try→Cys	TS
130	11	65	M	760	Sq	IB	7	244	GGC→TGC	G→T	−	−	Gly→Cys	TV
131	3	70	F	1000	Sq	IIIB	7	245	GGC→TGC	G→T	−	−	Gly→Cys	TV
132	5	79	M	2205	Sq	IB	7	245	GGC→CGC	G→C	−	−	Gly→Arg	TV
133	10	66	M	709	Sq	IIIA	7	245	GGC→CGC	G→C	−	−	Gly→Arg	TV
134	23	68	M	1440	Sq	IIIA	7	245	GGC→GTC	G→T	−	−	Gly→Val	TV
135	29	43	M	440	Sq	IIIB	7	248	CGG→CTG	G→T	−	−	Arg→Leu	TV
136	43	76	M	780	Sq	IIIB	7	248	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
137	46	81	M	1200	Sq	IIB	7	248	CGG→CTG	G→T	−	−	Arg→Leu	TV
138	12	72	M	705	Sq	IA	7	249	AGG→ATG	G→T	−	−	Arg→Met	TV
139	25	71	M	1020	Sq	IIIB	7	249	AGG→AGT	G→T	−	Arg→Ser	TV
140	34	74	M	2750	Sq	IIA	8	266	GGA→TGA	G→T	−	−	Gly→stop	TV
141	22	53	M	1320	Sq	IIB	8	271	GAG→TAG	G→T	−	−	Glu→stop	TV
142	39	65	M	760	Sq	IIIB	8	272	GTG→TTG	G→T	−	−	Val→Leu	TV
143	13	68	M	1950	Sq	IIIA	8	273	CGT→TGT	C→T	G→A (CpG)	C	Arg→Cys	TS
144	32	65	F	400	Sq	IIB	8	280	AGA→GGA	A→G	−	−	Arg→Gly	TS
145	8	59	M	1260	Sq	IIIB	8	282	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
146	20	75	F	0	Sq	IIB	8	282	CGG→TGG	C→T	G→A (CpG)	G	Arg→Trp	TS
147	1	61	M	0	Sq	IIIB	10	337	CGC→CTC	G→T	−	−	Arg→Leu	TV
148	4	76	M	600	Sq	IB	10	342	CGA→TGA	C→T	G→A (CpG)	G	Arg→stop	TS
149	17	51	M	1240	Sq	IIA	10	5′ junction	agAT→tgAT	a→t	−	−	Splicing	TV

^† Smoking index is the number of cigarettes/day × years. TS, transition; TV, transversion.

Figure 1.

Distribution of all point mutations (except junctional mutations) along the p53 codons in resected Japanese non‐small cell lung cancers (NSCLCs) by smoking status (a), and by histology (b): x axis, codon position; y axis, percentage of p53 mutations among all cases of each group. The labeled peaks indicate codon numbers, in which there were more than three mutations.

Relationships between p53 mutations and clinicopathological parameters.  The frequency of p53 mutations for smokers in all cases was 54%, which was higher than those for non‐smokers (37%) with statistical significance (P < 0.01) (Table 1). When compared by an amount of cigarette consumption, the frequency for heavy smokers was higher than that for light smokers with almost borderline significant difference (P = 0.057). Numbers of point mutations along p53 codons except at exon‐intron junctions were 50 for smokers and 25 for non‐smokers, 11 being found in common (Fig. 1a). Comparing the frequencies for the six hot spots, the sum for smokers (35%, 30/86) was higher than that for non‐smokers (23%, 7/31), though no statistical significance was observed (P = 0.21).

When p53 mutation frequencies were compared between histologies, the SQCC value was higher than for the ADCs (P < 0.01) (Table 1). Numbers of mutated codons were 30 for SQCCs and 48 for ADCs with 13 codons involved in common (Fig. 1b; Table 2). Mutations were most frequent in codon 220 for SQCCs, and codons 273 for ADCs, the difference for codon 220 being significant (P = 0.004).

Mean ages of the mutated cases were higher for SQCCs than for ADCs (P < 0.001) (Table 1). By gender, the mutation frequency was higher in males than females overall (P < 0.01). Regarding the pathological stages, the number with p‐stage IV was small, so they were combined with those of p‐stage III (Table 1). For p‐stages III + IV, percentages of mutated cases were 52% for ADCs, 65% for SQCCs, and 55% overall. Frequencies of mutations increased with the advance of p‐stages for all cases and for ADCs (P = 0.015 and 0.017, respectively).

p53 mutational spectra ( Table 3 ).  Most p53 mutations were transitions (61/297, 21%) or transversions (61/297, 21%), and deletions/insertions accounted for only 9% (27/297). The frequency of transversions in the smokers was higher than in non‐smokers with statistically significant difference (P < 0.001). By histology, frequencies of the former two were higher in SQCCs than in ADCs with statistical significance (P < 0.05 and < 0.01, respectively), but no difference was evident for deletions/insertions.

Table 3.

p53 mutation spectra by smoking status and by histology, and strand bias

With regard to base substitutions, G→T transversions accounted for 40 out of 297 total cases (13%). By smoking status, the frequency in the smokers was 18%, which was higher than that of non‐smokers (6%) with significance (P = 0.003), although there was no difference between histologies (19% in SQCCs and 12% in ADCs). Distributions of G→T transversions along the p53 codons by smoking status and by histology are shown in Fig. 2(a,b). The total number of mutations was 38, as two mutations observed in intron‐exon junctions were excluded. By smoking status, there were 32 mutations in 24 codons for smokers, and six in five codons for non‐smokers, and only two codons were commonly mutated. For smokers, all five hot spot codons for adduct formation by PAHs (exclusion of codon 249 from the six lung cancer hot spot codons) were included, and the average mutation number in each of the five codons was 2.0 (10/5); two times higher than in the other codons, 1.1 (22/19). In non‐smokers, mutations were observed in only two out of the five hot spot codons, and the average mutation number was 0.6 (3/5), lower than for other codons, 1.0 (3/3) (Fig. 2a; Table 2).

Figure 2.

Distribution of G→T transversions along p53 codons by smoking status (a), and by histology (b): x axis, codon positions; y axis, percentages of G→T mutations among all cases of each group. All hot spots (codons 157, 158, 245, 248 and 273) for adduct formation by polycyclic aromatic hydrocarbons (PAHs) were mutated in smokers (a). The mutations were distributed more widely along p53 codons for adenocarcinomas than for squamous cell carcinomas (b).

The distribution patterns also differed with the histology. Mutations numbered 14 over 11 codons for SQCCs, and 24 over 20 codons for ADCs, with only four commonly mutated codons (Fig. 2b; Table 2). Average mutation numbers for the five hot spot codons was 1.0 (5/5) and almost the same as for other codons, 1.1 (9/8), for SQCCs. For ADCs, the figure for the five hot spots 1.6 (8/5) was higher than for the others, 1.0 (16/16).

CpG→CpA transitions were observed in 10% of all cases, and mutation frequencies were equal across smoking status and between histology types. This base substitution occurred in a 5′‐CGT‐3′ sequence context on a non‐transcribed strand of two smokers (DNA nos. 50 and 174) and two non‐smokers (DNA nos. 49 and 83), all of which had ADCs (Table 2).

A→G transitions were observed in 6% of all cases, and the frequencies were the same for smokers and non‐smokers, but higher in SQCCs than in ADCs (P < 0.01).

Silent mutations of p53 were observed in two ADCs of smokers, case DNA nos. 136 (ACG to ACT; Thr to Thr) and 363 (ACG to ACA; Thr to Thr). Neither of these cases had any other p53 mutations.

Strand bias ( Table 3 ).  A marked strand bias was observed for G→T transversions, 98% (39/40) occurring on a non‐transcribed strand. Other transversion mutations, G→C, A→T and A→C, also showed a bias towards the non‐transcribed strand (78%, 67% and 67%, respectively). In addition 83% of A→G transitions were on non‐transcribed strands. In contrast, the CpG→CpA transitions were equally distributed on both strands (14 vs 16).

Discussion

The frequency of p53 mutations for NSCLCs (SQCCs and ADCs) in this study was 48%, approximately the same as in our previous study (45%) in 151 cases and other Japanese studies (48% each).^{( 20} , ²² , ^{23 )} Most mutations were found in the DNA binding domain of the gene, as described earlier.^{( 24 )} The frequencies by histological types and by gender, 62% in SQCCs and 43% in ADCs, and 54% in male and 37% in female, were also similar to the outcomes in past Japanese studies (58–67% vs 35–41%, respectively, for histology, and 50–57% vs 23–36%, respectively, for gender).^{( 20} , ²² , ^{23 )} Thus, our examined cases can be considered representative of lung cancer series in Japanese.

The p53 mutation frequencies in smokers were obviously higher than those in non‐smokers in our cases. When compared by tobacco consumption, a trend for more frequent mutations in heavy than in light smokers was also recognized, in line with other studies.^{( 6} , ^{22 )} Further, distribution patterns of the mutations along the p53 codons differed with the smoking status. Thus, p53 mutations are strongly related to tobacco smoking as reported.^{( 25} , ^{26 )}

Metabolites of polycyclic aromatic hydrocarbons (PAHs) and tobacco‐specific N‐nitrosamine (4‐methylnitrosamino‐1‐(3‐pyridyl)‐1‐butanone (NNK) included in tobacco smoke, and several reactive lipid peroxidation products derived from tobacco smoke, lead to the formation of DNA adducts and cause G→T transversions.^{( 27} , ²⁸ , ^{29 )} In our study, G→T transversions were observed more frequently in smokers than in non‐smokers with significance (P = 0.003), which was in line with previous reports.^{( 8} , ³⁰ , ^{31 )} Although Paschke insisted that he could not find significant differences between smokers and non‐smokers in frequencies of the mutations on analysis of the R3 version of IARC TP53 database, this is an exceptional report.^{( 32 )} Thus we conclude that the frequency of G→T transversions is higher in smokers than in non‐smokers.

The bulky adducts formation with tobacco carcinogenic chemicals at mutational hotspots of the p53 gene, especially in codons 157, 158, 245, 248, and 273, results predominantly in G→T transversions.^{( 13} , ^{26 )} In our study, such mutations were observed in the five codons for smokers, but only two codons for non‐smokers. Furthermore, in smokers, the average incidences at each of the five hot spots was two times higher than in others, and the results were reverse in non‐smokers, although the numbers were small. Similar results have been reported.^{( 24} , ^{31 )} In contrast, Rodin claimed that the G→T transversion distribution pattern along the p53 gene was indistinguishable between lung cancers of smokers and non‐lung cancers, and most likely not associated with smoke.^{( 14 )} However, even in Rodin's analyzed cases, frequencies of the mutations on the five codons were clearly higher in smokers than in non‐smokers. The importance of differences in mutation frequency should be stressed to distinguish lung cancers in smokers from non‐smokers and cancers ‘less accessible to smoke’, because lungs of non‐smokers and tissues of non‐lung cancers may also be either exposed to certain levels of PAHs directly or PAH metabolites indirectly, so the patterns may look somewhat similar to those in lung cancer, if one examines only G→T events.^{( 8 )} Thus, it is clear that there is a relatively precise correspondence between the mutational hot spots of lung cancer and the hot spots of adduct formation by carcinogens found in tobacco smoke.^{( 33} , ^{34 )}

In the present study, G→T transversions showed a marked strand bias; 98% were found in a non‐transcribed strand, which is in line with previous reports.^{( 3} , ^{5 )} This phenomenon has been considered to be one of the characteristics of mutations caused by DNA adducts derived from exogenous carcinogens, such as smoke mutagens.^{( 5} , ³⁵ , ³⁶ , ³⁷ , ^{38 )}

Reactive oxygen species (ROS), produced by endogenous mechanisms, such as cell aerobic metabolism and inflammation, or metabolism of PAHs to o‐quinone, are another source of G→T mutations. However, so far, it is uncertain how endogenous mechanisms contribute to the lung cancer specific mutation spectrum, with G→T transversions on non‐transcribed strands.^{( 29} , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ^{43 )}

Frequencies of the CpG→CpA transition have not been as extensively studied as G→T transversions. Here, they were found in 10% each in smokers and non‐smokers. Such transitions have generally been considered due to elevated susceptibility to spontaneous deamination. However, recently, in vitro evidence was presented indicating that the mutations may originate through adducts formation with metabolites of exogenous agents. Thus the (+)‐anti diol epoxide of BaP gave a preponderance of G→A mutations in a 5′‐CGT‐3′ sequence context,^{( 15 )} and these mutations are likely to be attributable to the major adduct. We found four human lung cancers with such lesions, all of them involving non‐transcribed strands. However, as a whole, the mutations were observed evenly in both transcribed and non‐transcribed strands, indicating that bulky adduct formation by exogenous carcinogens such as those included in tobacco smoke, do not play an important role in their genesis.

Other mechanisms that cause G→A transitions at CpG sites, other than spontaneous deamination, may be deamination of 5‐methylcytosine by certain chemicals like nitric oxide, oxidative DNA base damage, and enzyme‐catalyzed events.^{( 44 )} Nitric oxide may be involved in the pathobiology of the airway inflammatory process in chronic obstructive pulmonary disease, but most of our examined cases did not have any history of chronic obstructive disease.^{( 45} , ^{46 )} Whether oxidative mechanisms and enzymatic reactions that are endogenous are significant events in vivo for G→A transitions at CpG sites is not clear at present. G→A transitions may also occur at guanine of non‐CpG sites through formation of O⁶–methylguanine as a consequence of exposure of DNA to tobacco‐specific nitrosamines such as NNK.^{( 31} , ^{47 )} However, at CpG sites, most cytosines are methylated, and methylated cytosine in p53 protects neighboring guanine from O⁶–alkylation by NNK,^{( 18 )} explaining the preferential occurrence of these adducts at non‐CpG sites. From this discussion, we consider that most GpG→CpA transitions in lung cancers are attributable to endogenous mechanisms.

Regarding A→G transitions, the tobacco‐related differences have been reported only in two papers, so far.^{( 6} , ^{31 )} However, in our study, the mutation frequencies did not show any differences between smokers and non‐smokers, although strong strand bias, 83% on non‐transcribed, was found to be described in some papers.^{( 3} , ⁶ , ^{31 )} This observation is an indicator of mutagenesis by exogenous chemical compounds in the environment throughout the formation of bulky adducts. The observed preferential existence of the mutations in squamous cell carcinomas compared to adenocarcinomas has not been documented before. The reasons of why are not clear, but it could be due to differences in carcinogen exposure between distinct mucosal sites, squamous cell carcinomas usually arising at large bronchi or more proximal portions of the bronchioles and adenocarcinomas with more distal bronchiolar portions, or variation in carcinogenic activation or DNA repair by progenitor cells of the histologies.

Ethnic characteristics.  There has been a paper reporting frequencies of the CpG→CpA transitions among all p53 mutations of 8% for smokers and 14% for non‐smokers in eastern countries, but 12% and 23%, respectively, in westerners.^{( 6 )} Our data calculated in the same way were more in line with the latter: 17% (19/109) and 28% (11/40), respectively,. This may also indicate that endogenous mechanisms may be playing a more important role in the genesis of NSCLCs than in other eastern countries.

Histological differences in p53 mutation spectra.  The fact that our p53 mutation frequency was higher in SQCCs than in ADCs is in line with other reports,^{( 3} , ^{22 )} this presumably being related to the higher smoking rate for the former than the latter.^{( 6} , ^{22 )} In addition, compared to ADCs, uniquely high p53 mutations on codon 220, which is not a hot spot for PAHs, and more frequent A→G transitions with strand bias, of which induction were not affected by smoking status, were observed in SQCCs. This indicates that genesis of SQCCs may be more related with both tobacco smoke compounds except PAHs and environmental bulky‐adduct forming carcinogenic agents. Whereas, in ADCs, G→T transversions were more widely dispersed along the p53 gene, which implies roles for other exogenous factors for the genesis of ADCs

We could not find differences between ADCs and SQCCs in frequencies of CpG→CpA transitions and the G→T transversions, which is in agreement with the data of Calvez.^{( 31 )} However, there was one report of less frequent CpG→CpA transitions for ADCs than SQCCs, and more common G→T transversions in ADCs.^{( 48 )} This discrepancy may partly be related to differences in frequencies of either histological subtypes of ADCs or locations of SQCCs by ethnicity or geography. In this context it should be stressed that we have reported that the frequencies of these mutations significantly differ among subtypes of ADCs or among locations of SQCCs.^{( 19} , ^{20 )}

Increase of p53 mutation frequency with progression of carcinoma.  Adenocarcinomas at advanced stages showed more frequent p53 mutation than those at early stages (stage I) with significant differences. To our knowledge, this is the first report to note an increase of mutation rates with advance in the stages in ADCs of the lung. This may be for the following reason. Usually patients are continuously exposed to carcinogenic agents that can induce p53 mutation, even after their carcinoma development, so that cancer cells without p53 mutation at an early development stage remain at risk of mutation of the gene. If these mutations occur in the cancer cells, the mutated cells might gain a growth advantage superior to other cancer cells, and this could account for the greatest volume of carcinomas.