© 1999 Cancer Research Campaign Article no. bjoc.1998.0234 Loss of heterozygosity at chromosome 1p in different solid human tumours: association with survival

The distal half of chromosome 1p was analysed with 15 polymorphic microsatellite markers in 683 human solid tumours at different locations. Loss of heterozygosity (LOH) was observed at least at one site in 369 cases or 54% of the tumours. LOHs detected ranged from 30–64%, depending on tumour location. The major results regarding LOH at different tumour locations were as follows: stomach, 20/38 (53%); colon and rectum, 60/109 (55%); lung, 38/63 (60%); breast, 145/238 (61%); endometrium, 18/25 (72%); ovary, 17/31 (55%); testis, 11/30 (37%); kidney, 22/73 (30%); thyroid, 4/14 (29%); and sarcomas, 9/14 (64%). High percentages of LOH were seen in the 1p36.3, 1p36.1, 1p35–p34.3, 1p32 and 1p31 regions, suggesting the presence of tumour-suppressor genes. All these regions on chromosome 1p show high LOH in more than one tumour type. However, distinct patterns of LOH were detected at different tumour locations. There was a significant separation of survival curves, with and without LOH at chromosome 1p, in the breast cancer patients. Multivariate analysis showed that LOH at 1p in breast tumours is a better indicator for prognosis than the other variables tested in our model, including nodal metastasis. © 1999 Cancer Research Campaign

The majority of invasive human solid tumours are considered to be sporadic. Accumulation of multiple genetic alterations play a major part in tumorigenesis of these tumours. Mutations cause some genes (proto-oncogenes) to gain function while other genes [tumour-suppressor genes (TSG)] sustain loss of function (for review see Haber and Harlow, 1997). The localization of these genes to specific chromosome regions is currently a major area of study in cancer research. Loss or inactivation of TSG has been shown to be a major feature in the genesis of the solid cancers examined so far (Cavenee and White, 1995). As a full inactivation of a gene usually requires the silencing of both its alleles, the first inactivating mutation is, by inference, recessive. This first mutation can be somatic or passed through the germ line. The second mutation is somatic and proceeds through a chromosomal mechanism, which leads to loss of the wild-type allele (loss of heterozygosity) or replacement of the wild type allele by the mutant allele. The result is a complete absence of the normal protein product.
Loss of heterozygosity (LOH) on the short arm of chromosome 1 in solid human tumours has been reported in the following cancer types: breast cancer, neuroblastoma, mesothelioma, melanoma, testis cancer, liver cancer, stomach cancer, phaeochromocytoma, thyroid cancer, meningioma, colorectal cancer, endometrial cancer and Wilms tumour (Mathew et al, 1987;Dracopoli et al, 1989;Chen et al, 1992;Bardi et al, 1993;Taguchi et al, 1993;Bello et al, 1994;Bieche et al, 1994;Mathew et al, 1994;Stock et al, 1994;Yeh et al, 1994;Caron et al, 1995;Kuroki et al, 1995;Munn et al 1995;White et al, 1995;Di Vinci et al, 1996;Ezaki et al, 1996;Ragnarsson et al, 1996;Ogunbiyi et al, 1997;Vargas et al, 1997;Arlt et al, 1996;Steenman et al, 1997). All of these studies strongly indicate that TSGs may be located on the short arm of chromosome 1. A number of studies have shown a significant association between LOH at 1p and prognostic factors. The distal half of chromosome 1p from 1p31.1 to 1p36.3 is among the regions on 1p that have shown frequent LOH. At least two studies have shown an association between LOH at 1p and amplification of the N-myc proto-oncogene in breast tumours and neuroblastoma (Bieche et al, 1994;Caron et al, 1995). There seems to be an association between LOH at 1p and poor prognosis in patients with neuroblastoma (Caron et al, 1996), breast cancer (Ragnarsson et al, 1996) and colon cancer (Ogunbiyi et al, 1997). LOH at 1p was shown to be an early event in the carcinogenesis of breast cancer (Munn et al, 1995), liver cancer (Kuroki et al, 1995) and colorectal cancer (Di Vinci et al, 1996). One study indicates that LOH at 1p is a late event in melanomas (Dracopoli et al, 1989). Introduction of chromosome 1p36 into colon cancer cell line suppresses tumorigenic behaviour (Tanaka et al, 1993).
In this study we made an attempt to discover whether LOH patterns were similar in different types of human solid cancers. We screened 682 human solid tumours from 20 different locations for loss of heterozygosity at chromosome 1p, using 15 highly polymorphic microsatellite markers focusing on 1p31-pter. Furthermore, we tested whether there was an association between LOH at 1p and overall patient survival and clinico-pathological variables.

DNA extraction and analysis
Tumour and normal DNA was extracted from the tissue samples with proteinase K using a method developed for paraffin-embedded tissue . In the case of the breast samples, normal DNA was extracted from peripheral blood leukocytes by a saltingout method (Miller et al, 1988) and tumour DNA was extracted according to standard protocols. Paired blood and tumour DNA was subjected to PCR analysis. DynaZyme TM polymerase (from Finnzymes Oy, Espoo, Finland) was used in the buffer solution provided by the manufacturer. Samples were subjected to 35 cycles of amplification, consisting of 30 s at 94°C, 40 s at 55°C and 30 s at 72°C, followed by 10 min at 72°C. The markers used (Table 1) were obtained from Research Genetics (Huntsville, AL, USA). Primers were elongated by terminal-transferase in 40 mM K-HEPES/1 mM CoCl 2 buffer at pH 7.2 and 37°C over night. PCR products were separated on 6.5% polyacrylamide 8 M urea sequencing gels and transferred to a Hybond-N + nylon film (Amersham, Aylesbury, UK). They were then hybridized for at least 2 h at 42°C with the elongated primers, covalently labelled with peroxidase (ECL kit, Amersham, Aylesbury, UK). The membranes were washed once in 3 × SSC/0.1% SDS at 39-42°C and then twice in 0.2 × SSC at 39-42°C. After washing, the membranes were bathed in a detection reagent containing H 2 O 2 , luminol and an enhancer (ECL kit, Amersham, Aylesbury, UK) for 1 min at room temperature and signals were detected on DUPONT Cronex-4 film. Any absence or significant decrease (more than 50%) in the intensity of one allele relative to the other was considered LOH (Figure 1).

Statistical analysis
A chi-squared test was used to assess the relationship between LOH at 1p and prognostic variables. In the breast tumours LOH at 1p was compared with node status, tumour size, histological type, age of diagnosis, steroid receptor content, S-phase fraction and tumour ploidy. In the other tumours LOH at 1p was compared with node status, tumour size and tumour differentiation. Survival curves were calculated according to the method of Kaplan and Meier (Kaplan and Meier, 1958). Tests of difference between curves were made with the log-rank test for censored survival data (Mantel, 1960). Multivariate analysis was performed with Cox's partially nonparametric regression model (Cox, 1972). The Survival Tools for Statview Package (Abacus Concepts, Inc., Berkeley, CA, USA) was used for the statistical analysis.

Figure 1
The picture demonstrates a trinucleotide repeat polymorphism in five matched normal (N) and tumour (T) tissues from breast cancer patients. The microsatellite marker D1S488 was used in PCR analysis and the amplified products separated by electrophoresis in a 6.5% polyacrylamide 8 M urea denaturing gel. Case numbers are shown at the top. Numbers to the right indicate the size (in nucleotides) of the PCR product. Deletion can be seen in tumours 982 and 1019. The fact that there is not a complete loss of allele in these tumours is most probably due to contamination from normal DNA in the tumour sample. The tumours 981 and 983 have normal heterozygous allele patterns and patient sample 1020 is a homozygote

Association with clinical and pathological variables
Chi-squared analysis comparing tumours with LOH at chromosome 1p with clinical and pathological variables showed a weak association between LOH at 1p and high S-phase fraction (P = 0.021) in the breast tumours. There was also an association between LOH at 1p and poor tumour differentiation (grade III) in the colorectal cancer cases (P = 0.047), but association was not found in other cancers (lung, stomach, endometrium, ovary, kidney or testis). There was no significant association between LOH at 1p and node status or tumour size in any cancer types (breast, lung, colorectum, stomach, endometrium, ovary, kidney or testis). There was no association between LOH at 1p in the breast tumours and parameters such as histological type, age at diagnosis, ploidy, or steroid receptor content. Furthermore, there was no association between LOH at 1p and Duke's grade in colorectal cancer cases (data not shown).

Survival analysis
Breast cancer patients showed significant association between LOH at chromosome 1p and overall survival as tested with a logrank test (P < 0.001). There was a significant separation of survival curves ( Figure 5A). Median follow-up time is 5.0 years. When individual markers were analysed separately, with respect to association with breast cancer patients' overall survival, only one marker showed significance with a long-rank test (P = 0.022) -the D1S435 marker, located at the chromosome region 1p31.1. Colorectal cancer patients showed a trend between LOH at 1p and survival (P = 0.085) but the median follow-up time was only 2.1 years. Figure 5A shows the graphic representation of the survival statistics in the breast cancer patients and Figure 5B shows the survival statistics in colorectal cancer patients. Only 66 of the 146 breast cancer patients with LOH at 1p had metastasis in lymph nodes but 80 were lymph node-negative. Multivariate analysis was undertaken to evaluate the possible clinical relevance of LOH at 1p as a prognostic factor in breast cancer patients. The analysis showed that breast cancer patients with tumours with LOH at 1p had nearly a three-fold increase in relative mortality rate compared with patients with tumours without LOH at 1p. The RR (relative risk of dying in the multivariate analysis) was 2.7 (95% confidence interval: 1.5-4.9) and P < 0.001. The only other factor of prognostic value in the multivariate analysis was axillary nodal involvement with RR = 1.8 (95% CI: 1.1-3.0) and P = 0.015 (Table 2A). A multivariate analysis in colorectal cancer patients showed that patients with LOH at 1p had a nearly twofold increase in relative mortality rate compared with patients with tumours without LOH at 1p. The 95% confidence interval of the multivariate analysis was 0.99-3.6 and P = 0.055 (Table 2B).

DISCUSSION
In this study LOH at chromosome 1p was detected with at least one marker in 54% of the tumours examined. The elevated frequency of LOH with certain markers suggests five regions of deletion indicating locations of putative TSGs. Known genes, located in these regions, that could possibly play a role in tumorigenesis are shown in Figure 2. The p53 homologue p73 on 1p36.33 (Kaghad et al, 1997) is one of these genes. A modifier gene of colon tumorigenesis (Mom1) in the mouse, which encodes secretory type II phospholipase A 2 , affects the numbers of polyps developing as the consequence of a mutation in the Apc gene (Min mouse). A human homologue of the candidate for the Mom1 locus has been mapped to 1p35-p36.1 (Praml et al, 1995). Other genes or homologues to genes involved in cell growth and fidelity are: transcription factors, E2F2 on 1p36.11 (Lees et al, 1993), PAX7 on 1p36.1 (Vorobyov et al, 1997) and L-myc on 1p34.3 (Speleman et al, 1996); replication protein gene RPA2 on 1p35 (Ozawa et al, 1993); death receptor genes, TR2 (Kwon et al, 1997), TNFR2 (Kemper et al, 1991), DR3 (Bodmer et al, 1997) and DR5 (Wu et al, 1997), all on 1p36; other receptor genes, PTAFR on 1p34.3-p35 (Chase et al, 1996), TGFβR3 on 1p32-p33 (Johnson et al, 1995) and IL12Rβ2 on 1p31.2 (Yamamoto et al, 1997); prostaglandin receptors, PTGER3 on 1p31.2 and PTGFR on 1p31.1 (Duncan et al, 1995); tyrosine kinases, ECK on 1p36.1 (Sulman et al, 1997), LCK on 1p35 (Volpi et al, 1994) and JAK1 on 1p31.3-p32.3 (Modi et al, 1995); dual specific kinase ERK (Saito et al, 1995) on 1p36.1; repair protein genes, RAD54 on 1p32 (Rasio et al, 1997); and GADD45 on 1p31.1-p31.2 (Hey et al, 1996); cell cycle control proteins ,   70  60  50  40  30  20  10  0  D1S243  D1S468  D1S214  D1S228  D1S507  D1S436  D1S201  D1S209  D1S216  D1S207  D1S488  D1S167  D1S435   70  60  50  40  30  20 10 0  CDC2L1 (Lahti et al, 1994) on 1p36.2, RIZ on 1p36.1 (Buyse et al, 1996) and p18 (INK4C) on 1p32 (Lapointe et al, 1996). A difference in LOH was detected between different tumour types. Some tumour types, testis and renal cancer, show low frequency of LOH while other tumour types have a high frequency of LOH, the highest in sarcomas and endometrial cancers. A subset of markers show elevated frequency of LOH in a number of different tumour types, suggesting that loss of a given TSG might be involved in pathogenesis of tumour growth in different tissues. Otherwise, a distinct pattern of LOH was detected in individual tumour types, reflecting multiple TSGs, that could be differentially inactivated in tumorigenesis. Furthermore, the study shows that LOH at 1p can be detected in tumours of the mouth, oesophagus, liver, pancreas, prostate, brain and adrenal gland, as well as melanoma and lymphoma.

A B
There was a weak significant association between LOH at 1p and high S-phase fraction in breast tumours and between LOH and poor tumour differentiation in colorectal tumours. In breast cancer patients, there seems to be no association between LOH at 1p and age of onset, tumour ploidy or steroid receptor content. Furthermore, no association was detected between LOH at 1p and tumour size or nodal metastasis in tumours of the breast, stomach, colorectum, lung, endometrium, ovary, kidney or testis. LOH at 1p does not appear to be associated with the nodal metastasis step in carcinogenesis. LOH at 1p is an independent prognostic factor for the breast cancer patients and is in line with our earlier finding with less patient material and a shorter follow-up time (Ragnarsson et al, 1996). Furthermore, LOH at 1p seems to be a prognostic factor in colorectal cancer also, though in this study the association is not significant probably due to the short follow-up time. In a different study it was concluded that allelic loss in the   Marker  D1S243  D1S468  D1S507  D1S436  D1S233  D1S496  D1S209  D1S216  D1S465  D1S207  D1S488  D1S167  D1S435   A   686  767  809  827  871  929  948  573  769  741  755  876  856  872  879  930  1033  1186  519  704  882  1277  997  984