A paradigm for oncogene complementation in human breast cancer

© INSTITUTPASTEUR/ELsEVIER Paris 1989

Res. Virol. 1989, 140, 571-591.

A P A R A D I G M FOR ONCOGENE COMPLEMENTATION IN H U M A N BREAST CANCER M. Roux-Dosseto and P.M. Martin URA CNRS 1175, Facultd de M~decine-Nord, Bd P.-Dramard, 13326 Marseille Cedex 15 (France)

SUMMARY

Multiple cellular oncogenes are amplified in malignant tumours, and it is possible to invoke gene dosage as a possible base for product activity. In vitro data have shown that two different oncogenes can cooperate in converting a normal cell into one that is tumorigenic. This suggested that multiple cooperative alterations might be involved in cancer progression. Breast cancers have a broad spectrum of clinical behaviours ranging from highly agressive neoplasms to almost chronic diseases. Fifty months of clinical follow-up were studied in 143 patients with primary breast cancers. Uni- and multivariate analyses were performed in order to determine any synergistic effect of amplified c-rnyc, erbB-2 and int-2 genes on the disease-free state and overall survival. We showed that c-myc amplification was associated with early recurrence and shorter survival; in contrast, erbB-2 and int-2 extra copies resulted in later relapse events, especially in patients whose tumours showed a normal copy number of c-myc genes. This pointed out sequential activation of complex regulatory cascades within the cell. Such particular behaviour enabled us to categorize erbB-2 and int-2 oncogenes into a group showing delayed action, in contrast to c-myc involvment in rapid spread of the turnout. As expected, co-amplified c-myc and erbB-2 genes showed positive cooperation with respect to recurrence and shortening of overall survival. Finally, the harmful effects of amplified c-myc and erbB-2 oncogenes were dramatically increased in patient subgroups showing a normal copy number of the int-2 gene. Multivariate analysis was used to evaluate disease-free survival and to test for potential interactions of oncogene covariates. It pointed out multiple

Submitted August 10, 1989, accepted November 3, 1989.

572

M. R O U X - D O S S E T O

AND P.M. MARTIN

independent combinations which enabled us to define complementation groups with respect to clinical patient behaviour. KEY-WORDS: Oncogene, Amplification, Complementation, Tumour; Paradigm, Breast cancer, Prognosis.

INTRODUCTION The multistep nature of carcinogenesis is a widely accepted concept. Each step toward the tumorigenic state may reflect the activation, mutation or loss of genes. Since pioneering data showing that DNA tumour viruses such as polyoma and adenovirus carry at least two genes acting in a coordinated manner in primary culture of rodent cells (Houweling et al., 1980; Rassoulzadegan et al., 1982; Van den Elsen et al., 1982), Land et al. (1983) and Ruley (1983) devised a system to analyse the function of oncogenes. This has led to a number of important findings concerning oncogene cooperation (Land et al., 1984). Essentially, it has been shown that oncogenes may be divided into two classes, those involved in immortalizing functions and those involved in the transformed phenotype. Nevertheless, these experiments cannot carefully assess the host influences presumed to be critical in the emergence and progression of malignancies in vivo. Transgenic mice expressing foreign genes introduced into their genome enable testing of the efficiency of oncogenes in tumorigenesis. The transforming potential of an activated oncogene was found to be a function of the differentiated state of cell in which it was activated, as well as the cell type (Sinn et al., 1987 ; Andres et al., 1987 ; Schoenenberger et al., 1988). In addition, tumour formation developed more rapidly in F I dual carriers than in transgenic parents expressing only one activated oncogene, and oncogene cooperation occurred in vivo to enhance tumour development (Sinn et al., 1987). Finally, the spectrum of developed tumours was found to reflect the intrinsic oncogenic potential of some oncogenes in a particular context, rather than tissue-restricted expression. This suggested that other cooperating alterations might be prevalent, depending on the tissue (Sinn et al., 1987; Compere et al., 1989). The purpose of this report is to present data allowing a better understanding of oncogene complementation in the progression of human breast tumours. H u m a n breast cancer stands out for its broad spectrum of clinical behaviour, ranging from that of a highly aggressive neoplasm which disseminates shortly after diagnosis and causes death rapidly, to that of a vir-

DF ER

= =

disease-free (status). estrogen receptor.

PgR OS

= =

progesterone receptor. overall survival.

ONCOGENE COMPLEMENTATION IN BREAST CANCER

573

tualIy chronic disease, compatible with m a n y decades o f survival. In highly invasive tumours, metastases are found in half o f the patients during the first 30 m o n t h s after surgical treatment. In this study, only early (0-30 months) and intermediate (30-50 months) events, which are the critical periods for t u m o u r invasiveness, were considered. The term "disease-free" (DF) was used for the length o f the interval f r o m surgery to relapse (considered to be essentially local recurrence, regional or distant metastases), whereas deaths due to breast cancer defined overall survival (OS) o f patients. A n u m b e r o f oncogenes and activation mechanisms have been shown to be involved in h u m a n m a m m a r y tumours. Valuable prognostic significance has already been assigned separately to c-myc, erbB-2 and int-2 gene amplification in malignant breast lesions (Escot et al., 1986; Slamon et aL, 1989; Lidereau et al., 1988). Here we have investigated the c o m p l e m e n t a t i o n effects o f i n t r a t u m o r a l c-myc, erbB-2 and int-2 gene status on DF and OS o f patients with p r i m a r y breast cancer.

MATERIALS AND METHODS Patients.

Breast tumour samples in liquid nitrogen were collected in our laboratory for routine assay of estrogen (ER) and progesterone (PgR) receptors and stored at - 80°C as part of our tumour library. From these, 170 primary breast carcinomas were retrospectively selected. Hospital records were reviewed and complete 60-month follow-up was available for 143 patients. The characteristics of the population entered in this study have been detailed elsewhere (Roux-Dosseto et al., 1989). Essentially, estrogen and progesterone receptors were assayed in our laboratory. Specimens were considered ER- and PgR-positive if they contained at least ten fmol of specific binding sites/mg of cytosolic protein. The diameter of primary tumours was assessed of cytosolic protein. The diameter of primary tumours was assessed by histologic measurements. Grading was performed according to Bloom and Richardsons classification (1957). Node involvement was estimated after surgery by careful examination of a minimum of ten nodes per patient. Half of the patients had no axillary node involvement and did not receive adjuvant therapy. In the remaining group (with involved nodes), no predominant therapy was performed and protocols were equally distributed within this patient subset which would preclude any selective bias due to treatments. Table I compares the characteristics of patients included in the study with a randomized population from the same location (MarseiUe area). D N A extraction and Southern blot analysis.

For all patients included in this study, samples of primary breast tumour were collected in liquid nitrogen and maintained at - 80 ° C until use. Total genomic DNA was prepared as previously reported (Roux-Dosseto et al., 1983). Intratumoral amplification of c-myc, erbB-2 and int-2 genes was estimated by Southern blot analysis. Ten micrograms of each genomic DNA sample were restricted with EcoRI or HindIII before electrophoresis and transfer to "Hybond-N" (Amersham). Prehybridizations, hybridizations and washes were performed as described (Brunet et al., 1986). Blots were further exposed for 18 to 48 hr on Amersham "Hyperfilm M P " at - 80°C with intensifying screens. The intensity of the bands was estimated

574

M. ROUX-DOSSETO

AND P.M. MAR TIN

TABLE I. - - Characteristics o f patients entering this study.

Study n = 170

Randomized n = 1 300

Nodal status =0 54.3°70 >0 45.7070 ER status < 10 35.8070 />10 64.2070

26.5070 73.5070

Size o f tumours (mm) < 20 39.7070 20-50 54.6070 /> 50 5.707o

56.6070 36.2o7o 4.2o7o

62.5070 37.5070

Study n = 170 <3 69.3070 />3 30.707o PgR status < 10 41.7070 i>10 58.3070 SBR grade I 18.4070 II 53.0o70 III 28.6070

Randomized n = 1 300 77.7070 22.3070 32.9070 67.1070 11.0% 58.007o 31.0070

by densitometric scanning of the autoradiograms (Shimadzu, CS 930). All experiments were performed without any knowledge of clinical data. Three levels of controls were performed: first, the amount of membraneimmobilized DNA actually available for hybridization was estimated using a single copy probe ([3-globin). Next, amplification was distinguished from aneuploidy by further rehybridization of blots with single copy probes mapped to the chromosomes in question. Finally, normal single copy signals were obtained from genomic DNA prepared from peripheral blood lymphocytes of healthy donors and analysed concomitantly. Tumour DNA samples showing signals with an intensity of two copies or more per haploid genome were considered to represent amplification. All experiments were repeated at least twice for each tumour DNA sample. The following oncogene probes were used: c-myc (1.5 kilobase, ClaI-EcoRI fragment; Alitalo et al., 1983) mapped to chromosome band 8q24; erbB-2 (0.440 kilobase, KpnI-XbaI fragment; Semba et al., 1985 and 3 kilobases, HindlII-KpnI fragment; Yamamoto et aL, 1986) mapped to chromosome band 17q21 (Schechter et al., 1985); int-2 (0.9 kilobase, SacI fragment), mapped to chromosome band 1 lq13 (Casey et al., 1986).

The control probes consisted of: c-mos (2.7 kilobase, EcoRI fragment, Watson et al., 1982), mapped to chromosome band 8q22;

human p53 (1.7 kilobase, StuI-SphI fragment; Zakut-Houri et al., I985), mapped to chromosome band 17p13 (Isobe et al., 1986); human retinoic acid receptor (2-kb fragment; Petkovitch et al., 1987) mapped to chromosome band 17q21 (Mattei et al., 1988); [3-globin (1.4 kilobase EcoRI-HindlII fragment; Krainer et al., 1984), mapped to chromosome band l lq15. All probes were labelled to high specific activity using the random primer method (Feinberg and Vogelstein, 1983).

ONCOGENE COMPLEMENTA TION 1N BREAST CANCER

575

Statistical analysis. Statistical analyses were performed using the Clinical Data Management System (Medlog, Information Analysis Corporation, Mountain View, CA). All quantitative data deduced from experimental evaluations of intratumoral c-myc, erbB-2 and int-2 gene copy number were transformed into qualitative variables: i.e., 0 if the number of oncogene copy was 1, and 1 if the copy number was 2 or more. Additional variables were created to describe subgroups corresponding to heterogeneous combinations of oncogene couples in the same tumour. Monoparametric analysis curves were drawn by the Kaplan-Meier method (Kaplan and Meier, 1958) to assess the effects of intratumoral c-myc, erbB-2 and int-2 gene amplification on disease-free and overall survival. Actuarial curves were deduced from event frequencies over 30-months (early events) and 50-month (intermediate events) follow-ups, and statistical differences between survival curves were evaluated with the log rank test (Peto et al., 1977). The p value compared differences in disease-free and overall survival of patients with and without oncogene amplification; p values lower than 0.05 were considered to be significant. To determine the significance of oncogene status on the time of relapse and to test for potential interactions of covariates, the Cox proportional hazards model (Cox, 1972) was used. The assumption of log-linearity of covariates was tested according to Lawless (1982).

RESULTS

Intratumoral c-myc, erbB-2 and int-2 amplification evaluation. Amplification of c-myc, erbB-2 and int-2 genes was found to concern, respectively, 27, 28 and 17% of primary breast tumours. A typical experim e n t is presented in figure 1. The presence or absence of amplification in patient D N A was detected by internal comparison with normal individuals. Figure 1A shows a hybridization pattern, in the same filter, obtained with c-myc (on the left) and int-2 (on the right) probes which detected a 12-kb band and a 14-kb band, respectively; tumours with amplified sequences were labelled with a star. In all experiments, the a m o u n t of membrane-immobilized D N A actually available for hybridization was assessed using a single copy cDNA probe. In figure 1B, the filter on the left was hybridized with the c - m y c probe, dehybridized and probed (filter on right) with erbB-2 (middle 14-kb band) and ~-globin (major 8.2-kb band) sequences. Normalization of c - m y c and erbB-2 signals was performed according to the [3-globin control. As aneuploidy frequently occurs in developing tumours, aneuploidy versus amplification was ruled out by further hybridization o f blots with single copy control probes m a p p e d to the chromosomes in question. In figure 1C, blots on the left show representative t u m o u r D N A with amplification of c - m y c (top), erbB-2 (middle) and int-2 (bottom) genes. Further hybridization o f the same blots (on the right) with c-mos (top), p53 (middle) and [3-globin (bottom) probes detected single copy signals. Since low amplification levels (2 to 5 copies per haploid

576

M. ROUX-DOSSETO

AND

P.M. MARTIN

genome) m a y be difficult to directly assess, we designed serial dilution experiments to c o n f i r m the low copy n u m b e r o f amplified sequences identified during the first r o u n d o f hybridization. In figure 1D, the blot on the left presents signals obtained with 2-fold (5%tg) a n d 5-fold (2-Ezg) dilutions o f cm y c amplified t u m o u r D N A c o m p a r e d to the undiluted (10 p.g) n o r m a l control (N), indicating that signals showen with oncogene probes concern genuine amplification o f these sequences. The filter on the right shows the 14-kb band detected with the e r b B - 2 probe in 2-fold (5-1zg), 5-fold (2%tg), 10-fold (1 [zg) and 50-fold (0.1-~tg) dilutions o f t u m o u r D N A . H y b r i d i z a t i o n was p e r f o r m ed simultaneously with the single copy [3-globin probe (major 8.2-kb fragment) in order to control the a m o u n t o f D N A actually available. The control (10-~tg) on the right shows double bands corresponding to n o r m a l e r b B - 2 and [3-globin genes. Thus, 2-fold dilutions o f D N A samples showing an intensity equal to undiluted n o r m a l D N A were definitively considered to represent amplification. In figure 1D, t u m o u r s n ° 55 a n d n ° 249, previously scored as c - m y c - and e r b B - 2 - a m p l i f i e d , were finally excluded. All experiments were repeated at least twice for each t u m o u r s D N A . The correlation between oncogene status in the same t u m o u r was assessed by chi-square analyses. C - m y c was f o u n d to be significantly co-amplified with e r b B - 2 ( p = 0 . 0 0 3 ) and int-2 ( p = 0 . 0 5 ) genes.

FIG. 1. -- Evaluation of oncogene amplification & tumours o f patients with primary breast cancers. Ten micrograms of tumour sample (number at top) or normal peripheral blood cell (N) DNA were digested with HindlII and analysed by Southern blot. Only relevant portions of filters are presented; -HindlII fragments on left (in kb). A) Filter on left was hybridized with c-myc probe which detected at 12-kb fragment. Following hybridization, the probe was stripped off and the blot re-hybridized (right) with the int-2 probe showing a 14-kb band. Sample signals were compared to the control and amplified sequences were scored (*). B) Normalization of oncogene signals according to the amount of DNA loaded in each lane was performed by systematichybridization with a single-copy[3-globinprobe. Filter hybridized with the c-myc probe on the left was washed and re-hybridized (filter on the right) with erbB-2 and [3-globinprobes; 14-kb band : erbB-2 gene detected with pKX044 probe; 8.2-kb band: major fragment detected with the [3-globin probe. (*)= Amplified tumours. C) Single-copy probes mapped to relevant chromosomes were used to check for aneuploidy vs. genuine gene amplification. Filters on left show representative amplified tumours and normal control (N) hybridized with c-myc (top), erbB-2 (middle) and int-2 (bottom) probes. Filters on right present the patterns of hybridization obtained when re-probing with c-mos (top), p53 (middle) and [3-globin (bottom). D) Low copy number of putative amplified genes was assessed by serial dilutions. Filter on left: 2-fold (5-1zg) and 5-fold (2-1~g) dilutions of tumour DNA were hybridized to the cmyc probe and the signals were compared to that obtained with the undiluted (10-~.g) control (1'4). Filter on right: 2-fold (5-1~g), 5-fold (2-~tg), 10-fold (1-t~g) and 50-fold (0.5-1zg) dilutions of tumour DNA were hybn'dizedwith the erbB-2 (pKK044) probe and 14-kb signals were compared with the corresponding one in the undiluted (10-~tg) control (N); 8.2-kb fragment: [3-globin control used to monitor amount of DNA on filter (*)= Amplified tumours.

as~

k41 I~

M

~B

31e

)

3

k

~

~

~

57

401

S~ ¸s~

411

~

~

~

65

~

86

~

~

N

75*

o m

~

a

~

~

m

~

~

s

~I

~

~

~

~ID

~2~

C m

N

~

N

~

~

N

578

M. R O U X - D O S S E T O

AND P.M. MARTIN

Intrinsic effects o f intratumoral c-myc, erbB-2 and int-2 gene amplification on DF and OS. DF (i.e. the time to relapse) (fig. 2) and OS (i.e. the time for the t u m o u r to cause death) (fig. 3) were investigated in order to evaluate the weight o f oncogene amplification in breast cancer progression. T u m o u r s with c - m y c amplification were c o m p a r e d to those showing a n o r m a l copy n u m b e r o f this oncogene. We observed a strong relationship between this parameter and both early relapse (fig. 2A) and decreased overall survival (fig. 3A). Most events occurred over a 30-month period (fig. 2A, p < 0.0001 ; fig. 3A, p = 0.0036) and only 30% o f patients were free o f disease after a 50-month follow-up (fig. 2A, p =0.0023). Figure 2B plots disease-free survival curves in patients whose t u m o u r displayed an amplified erbB-2 gene as c o m p a r e d to those with n o r m a l copy n u m b e r o f this gene. N o relapse was significantly scored over the early interval (30 months) and recurrence t o o k longer (50 m o n t h s , p = 0.0163). Conversely, the corresponding overall survival curves presented in figure 3B were roughly superimposable on the disease-free ones, t h o u g h a shift to lower significance was observed (p = 0.0483). This accounts for the gap between endpoints in DF and OS, as death occurred later after recurrence o f the disease. In contrast, int-2 amplification was not associated with any early or later events and patient behaviour did not present significant deflection as c o m p a r e d to the control group without int-2 amplification according to D F (fig. 2C) and OS (data not shown). Multiparametric analysis with oncogene covariates revealed a significant p value associated only with c-myc amplification (p = 0.0003). The nonsignificant p values associated with erbB-2 and int-2 amplification m e a n t that these alterations had little predictive values when c - m y c status was k n o w n . It also implied that erbB-2 and int-2 amplification status were not independent parameters and that t u m o u r behaviour m a y have hinged on their interactions. Intratumoral c-myc and erbB-2 complementation. C - m y c and erbB-2 gene amplification status in the same t u m o u r was investigated simultaneously. The three groups corresponding to putative com-

FIG. 2. -- DF survival related to c-myc (A), erbB-2 (B) and int-2 (C) intratumoral gene amplification. Triangles represent percent of surviving patients in group with oncogene amplification. Full circles represent percent of surviving patients in control group with normal copy number of corresponding oncogene. After 30-month and 50-month follow-up, data were supported by cumulative p values evaluated with log rank test (NS = not significant). Number at risk : number of study participants.

ONCOGENE C O M P L E M E N T A T I O N IN B R E A S T CANCER A 90140. P E

70-

R

C E N T

60. 50" 40'

L

30amplified c-myc

20---

normal c-mye

10.

DF I

MO.'~FHS

0

NOal risk&

31

N°ae risk •

84

I

l

l

l

l

~

B

90 140

I

I

I

I

30

50

<0.0001

0.0023

10

P values

I

I

I._ i ----I

70 60 50

t~

40 3020 =

--

amplified ©rbB-2

---

normal ¢rbB-2

10OF i

i

i

i

i

I

N°at risk •

i

i

i

i

30

50

NS

0.11163

lO

N°a!

i

I

i

39

risk •

104

P vulues

C

I 901407060504030amplified int-2

20--

normal int-2

10.

DF ]

0 N°al

risk •

22

NOa! risk •

106

P valuc~

|

I0

i

!

|

I

30

J

|

I

I

50

NS

]

]

579

580

M. ROUX-DOSSETO

AND

A

P.M. MARTIN

!

90

•

---91

\ A

HI) •

P E R C E N T

7It • 60 50. 40311 20I0

--

amplifncd

--

normal (-my,:

~-myc

"

OS I

MoNrrHs

0

NOat

risk •

31

N°al

risk •

84

I

I

I

I

I

[

I

3(I

5O

O.OOqh

0 I)l)(} I

I0

P value,

I

I

I

1

I

B

L'I" . . . . . "*' 90-

I •

~ I..-*

,R0706050403020-

--

amplirictl

--

normal cfbB-2

etbB-2

10-

O$ I

t 10

NOat risk • N'at

risk •

P values

I

I

I

1 30

i

i

i

i

i

I

50

,39 104 NS

(I.II483

FIG. 3. - - Overall surival related to c-myc (A) and erbB-2 (B) intratumoral gene amplification. See legends, figure 2.

FIG. 4. - - DF survival related to co-amplified c-myc and erbB-2 (A), amplified c-myc/normal erbB-2 (B) and normal c-myc/amplified erbB-2 (C) vs. normal copy number c - m y c / e r b B - 2 controls.

See legend, figure 2.

ONCOGENE C O M P L E M E N T A T I O N IN BRF__.4ST CANCER A

1.0_

90" 80,

P E R C E N T

I_.) '-~..

70,

le

60-~

t~

L._

504030amplified

20.-

c-myc/zmplified

normal c-myc/normal

rrbB-2

¢rbB*2

10-

DF I

I

MONTHS

I

I

I

|

IO

NOal risk •

15

N°al

63

risk •

P valu¢~;

I

I

I

I

I

30

50

0.0004

(1,0030

I

I

B

90 -

•

|

L,

XI} -

"-I

70 / 60 50"

L

40 30. 20"

amplified

1

--

c-myc/normal

erbB-2

nol'mal c-myc/normal erbB-2

)0-

DF i

i

risk •

If)

N°a=

rink •

63

I

i

I0

(I N°Ul

I

P valuc~.

I

i

i

I

i

i

}()

50

l),()(125

(I.(1(17(I

I

C

~o-

L~ ,

7ox"

L _~_°

6050403020-

--normal --

L c-myc/amplified

erbB-2

normal c-myc/norrnil erbB-2

)0

DF i

i II}

N°at

rb;k&

N°al r i ~ k l P valuc~,

f

i

i

I

l

I

I

i

30

5O

NS

0.OO67

15 63

i

I

581

582

M. R O U X - D O S S E T O A N D P . M . M A R T I N

binations were compared the control consisting of patients whose tumours exhibited normal copy number of both c-myc and erbB-2 oncogenes. The first group corresponded to homogeneous tumours with co-amplified c-myc and erbB-2 genes; the two others represented heterogeneous tumours showing only one amplified oncogene, i.e., amplified c-myc with a normal copy number of the erbB-2 gene (amplified c - m y c / n o r m a l erbB-2), and conversely, a norma~ copy number of c-myc with amplified erbB-2 (normal c-myc/amplified erbB-2). Using monovariate analysis, we challenged the disease-free (fig. 4) and overall (fig. 5) survival in these groups to those observed in the control population.

I---0-

4

I ;

80 P E R C E N T

I .......

I

90

70 60 50 40 30

amplified c-myc/amplified erbB-2

20

--

normal c-myc/normal erbB-2

10

OS I

MONTHS

0

NOat risk •

15

NOat risk •

63

P values

I

10

I

I

I

|

I

I

I

I

30

50

(1.0154

(1 0(103

I

I

I

FIG. 5. -- Overall survival related to c-myc and erbB-2 co-amplification versus normal copy number c-mys/erbB-2 controls. See legend, figure 2.

Effects on DF.

Figure 4A presents disease-free curves of patients with intratumoral e-myc and erbB-2 co-amplification. According to the control group showing a normal copy number of both c-myc and erbB-2 oncogenes, most events occurred over the 0-30-month period (p = 0.0004) as previously mentioned (fig. 2A). Figure 4B shows the risk of recurrence for patients whose tumours showed amplified c-myc and normal copy number of the erbB-2 gene (amplified c-

ONCOGENE COMPLEMENTA TION I N BREAST CANCER

583

myc/normal erbB-2). Relapse occurred primarily during the early 0-30-month period; nevertheless the p value had lower significance (p = 0.0025) compared to its counterpart in figure 4A (p = 0.0004). This suggested that positive cooperation might lead to early recurrence with increased frequency when both oncogenes were amplified. Figure 4C presents DF curves when only amplified erbB-2 was detected in association with a normal copy number of c-myc. Again, the data strongly suggested that erbB-2 amplification had little or no effect during the first 30 months. In contrast, this genetic alteration was associated with events occurring in the intermediate period (30-50 months). The final course of the disease (at 50 months) was roughly comparable to that resulting from combinations consisting of amplified c-myc associations with extra (figure 4A) or normal (figure 4B) copy numbers of erbB-2 genes. In any case, c-myc amplification behaved as the dominant parameter for early relapse. These observations suggested multiple processes in which tumour progression resulted from different oncogene combinations, such as erbB-2 amplification in a c-myc normal population alternating with amplified c-myc in an erbB-2 normal group. In others words, sequential cooperation mechanisms may have accounted for malignant spread. Multivariate analyses established that oncogene associations (i.e. homogenous combinations with co-amplified c-myc/erbB-2 genes as well as heterogeneous ones with either amplified c-myc/normal erbB-2 or normal c-myc/amplified erbB-2), behaved independently with respect to relapse risks. The corresponding p values were 0.0159, 0.0055 and 0.0230, respectively. All three combinations defined distinct complementation subsets on the basis of clinical patient behaviour.

Effects on OS. In the subgroups of patients defined above, overall survival curves were plotted to assess how long the turnout took to cause death..No significant deflection of the curves was observed in heterogeneous combinations showing only amplified c-myc or erbB-2 oncogenes compared to control patients with a normal copy number of both c-myc and erbB-2 oncogenes (data not shown). Conservely, in the subgroup with co-amplified c-mycand erbB-2 oncogenes, death rates dramatically increased in early (0-30 months) or later (30-50 months) periods (p = 0.0145 and p = 0.0003, respectively; figure 5). Thus, positive cooperation between c-myc and erbBe2 oncogenes was suggested by the currently observed p values.

Intratumoral c-myc and int-2 gene complementation: effects on D F and OS.

The DF curves of patients with dissociated c-myc and' int-2 status were:" compared to the control group exhibiting no detectable amplification of either oncogene. In figure 6A, we compared recurrence in patients with intratumoral

584

M. ROUX-DOSSETO AND P.M. MARTIN

amplified c-myc associated with a normal copy of int-2 to control patients showing no genetic alterations of these oncogenes. Paradoxically, we observed that in the established absence of any int-2 amplification (normal int-2), c-myc amplification was correlated with unfavorable evolution of the disease. It, therefore, displayed a great predictive value compared to c-myc amplification alone (fig. 2A) or associated with erbB-2 (fig. 4A). Conversely, figure 6B presents relapse risks in patients whose tumours showed amplified int-2 and a normal copy number of the c-myc gene. In this case, int-2 amplification was associated with later (50 months) recurrence (p = 0.0002), since most events occurred beyond the 30-month period. These data were reminiscent of those reported in figure 4C, where an amplified erbB-2/normal c-myc combination was considered. Patients with c-myc and int-2 co-amplified genes did not behave differently from the control population (data not shown). However, due to the small number of individuals in these groups, we cannot definitively rule out statistical differences. Overall survival curves of patients with the amplified c-myc gene and normal int-2 status were compared to controls in figure 6C. The p values associated with early (19= 0.0008) or later (p < 0.0001) events display higher statistical significance compared to a similar association with amplified c-myc in figure 4A (13= 0.0036 and p = 0.0001, respectively). Multivariate analyses established that heterogeneous combinations of c-myc and int-2 oncogenes were independent predictive parameters for relapse at 50 months (amplified c - m y c / n o r m a l int-2, p < 0 . 0 0 0 1 ; normal c-myc/amplified int-2, p = 0.0049). Intratumoral erbB-2 and int-2 gene complementation: effects on DF and OS.

Similarly, we investigated the DF and OS status of patients according to their intratumoral erbB-2 and int-2 gene status. Only the group with amplified erbB-2 and normal int-2 behaved differently from the control group (i.e., normal copy number of both int-2 and erbB-2 genes). The results are presented in figure 7. It is noteworthy that erbB-2 amplification displayed higher risks of recurrence (figure 7A, p=0.0224 an p=0.0019) and death (figure 7B, p =0.0316), in contrast to the former behaviour associated with amplified erbB-2 genes in global studies which did not consider the int-2 status (figure 2B, p: NS and p = 0.0163; figure 3B, p = 0.0483). This is in good agreement with the data previously mentioned in figure 6A and 6C showing the harmful effect of c-myc amplification in the apparent absence of any int-2 genetic alterations. Some undescribed negative cooperation mechanisms might account for these observations. In contrast, the two other subsets, consisting of erbB-2

FIG. 6. -- DF survival (,4 and B) and OS (C) related to amplified c-myc/normal int-2 (A and C) and normal c-myc/amplified int-2 (B) vs. normal copy number c-myc/int-2 controls. See legend, figure 2.

ONCOGENE C O M P L E M E N T A T I O N IN B R E A S T CANCER A 90.

I

l I

80-

l,

P E

70-

R

C

60 •

E N T

50 ' 40 •

I

30 20-

--

amplified

c-myc/normal

---

normal c-myc/normal

I

inl-2

[&

inl-2

I|) -

OF

t

MONTHS

0

Neat risk •

21

N°al

70

risk •

!

!

t

l

l

i

t

!

30

I0

P valucs

t

[

t

50

<[).(X)OI


B

90 80

h

70 60 50 40 30

normal c-myc /amplified inl-2

20-

---

normal c-myc/normal int-2

10 -

OF i

i

i

i

i

I

tO Neat

risk&

II

Neat

risk•

70

!

i

i

30

i

I

i

t

50

0.0(}02

N$

P valuc~

i

C 90-

\.

80-

7060504030--amplified

20-

¢-myc /normal

- - - - normal ¢-myc/normal

in|-2

int-2

I0-

OS !

i

I0 N°al

risk•

21

N°al

risk •

"/0

P values

i

i

i

I

30

0.0008

I

I

i

i

50


i

585

586

M. ROUX-DOSSETO A N D P.M. M A R T I N A 90

- --•'i

80 P E

I

L1•

70-

---5.

R

C E N T

60 50

t.

4O 30 2O

--

amplified

- -

normal

etbB-2/normal erbB-2/normal

lnl-2 inl-2

10

OF i

MONTHS

0

I

I

i

10

risk •

24

NOat risk •

77

N°al

i

1

i

i

i

30

.L

i

0.0019

B

90-

I

50

0.0224

P values

i

\_--

i

8070-

A

6050 40 30 aacplified

20 --

normal

erbB-2/normal

int-2

eebB-2/notmal i n t - 2

10

OS !

0 N'at

risk •

24

N°at risk •

77

P

FIG. 7.

-

-

values

i

i

10

I

|

i

|

I

i

I

i

30

50

NS

0.0316

i

!

D F survival (A) and OS (B) related to amplified erbB-2/normal int-2 vs. normal copy number erbB-2/int-2 controls.

See legend, figure 2.

and int-2 co-amplified oncogenes (amplified erbB-2/amplified int-2) and amplified int-2 gene associated with erbB-2 normal copy number (amplified int-2/normal erbB-2), displayed non-significant p values (data not shown). Multivariate analyses, reported in table 2A, confirmed that only normal int-2 associated with amplified erbB-2 was informative. Nevertheless, all data concerning int-2/erbB-2 associations was analyzed regardless of intratumoral

ONCOGENE COMPLEMENTA TION IN BREAST CANCER

587

TABLE II. - - Multivariate survival analyses comparing DF survival to both erbB-2 and int-2 intratumoral gene amplification status.

A) in unselected (amplified or normaO c-myc population: amplified erbB-2/amplified int-2 amplified erbB-2/normal int-2 normal erbB-2/amplified int-2

NS 0.0316 NS

B) in normal c-myc population: amplified erbB-2 /amplified int-2 amplified erbB-2/normal int-2 normal erbB-2/amplified int-2

0.0071 0.0302 0.0090

c-myc status, and it can be hypothesized that the high risk of recurrence associated with the former combination might, to some extent, involve amplified c-myc genes. Thus, the effects of int-2 and erbB-2 co-amplification were investigated independently of c-myc status in the group of patients showing a normal copy number of c-myc oncogene. Table 2B presents relapse risks according to covariates consisting of int-2 and erbB-2 combinations. Highly significant p values were consistently found, establishing the intrinsic oncogenic potential of all three associations in the absence of c-myc amplification.

DISCUSSION

The present retrospective study was performed to evaluate the complementary effects of amplified c-myc, erbB-2 and int-2 on breast cancer progression. This approach fitted and in some aspects anticipated in vitro and animal models aimed at assessing the oncogenic potential of different oncogenes or combinations of oncogenes. Indeed, experimental approaches provide insights only into positive cooperation enhancing the overall efficiency of tumour induction. Follow-up after surgery of the primary tumour provides additional information, especially concerning recurrence. We thus distinguished two groups of tumours according to their oncogene (co-) amplification status. They consisted of tumours showing c-myc amplification with a normal copy number of erbB-2 and, conversely, tumours with erbB-2 or int-2 amplification and anormal copy number of the c-myc gene. Remarkably, c-myc amplification was the powerful parameter associated with early relapse (0-30 months) and death. In contrast, amplification of erbB-2 and int-2 genes had no effect over this period; their oncogenic potential developed later (30-50 months). According to the multistage model for tumour progression (Moolgavkar et al., 1986), these findings suggested that the different oncogene combinations may result in similar effects (i.e. recurrence)

588

M. R O U X - D O S S E T O A N D P.M. M A R T I N

but within different time spans. In longer development periods, additional events may also occur, stimulating the oncogenic potential associated with erbB-2 amplification. In any case, the notion of sequential cooperation introduced a notion of time into putative modulation of carcinogenic potential of some oncogenes. In addition, our data indicated that c-myc and erbB-2 have a positive cooperation effect, with overall survival significantly decreased in patients showing amplification of both genes in their primary tumour. This supports speculation that c-myc and the erbB-2 oncogenes play an intrinsic role in breast cancer progression. It also supports the current hypothesis that some oncogenes act synergistically to drive cellular transformation. Since our data showed that the harmful effects of c-myc a n d / o r erbB-2 amplification increased in the absence of apparent int-2 alterations, it is tempting to speculate that negative cooperation occurs with these combinations. The human int-2 gene was previously mapped to the q13 band of chromosome 11 (Casey et al., 1986), which has been shown to contain the progesterone receptor (Law et al., 1987) and the hstl gene (Yoshida et al., 1988). In other works, int-2 and hstl genes have been found to be amplified as a single amplicon unit (Yoshida et al., 1988), and it has been suggested that there may be a cluster of hstl- or int-2-related genes. Recently, it has been shown that inactivation of the p53 proto-oncogene renders the normal cell predisposed to transformation, whereas overexpression prevents transformation (Finlay et aL, 1989). The normal function of int-2 products, as well as of co-amplified hstl, is not yet known and no suppressor activity has been mapped to the int-2 region. Elsewhere, chromosome loss associated with duplication of the second copy might result in apparent diploidy (Canevee et al., 1986). Such genetic alterations may occur in breast tumours, and the study of organization of this region in breast cancer may be of interest. Since we considered overall oncogene complementation in tumours, we could not distinguish between molecular events occurring either in the same cells or, through tumour heterogeneity, in different cells. In such a case, a wider concept should be invoked to include cell-cell interactions. Investigation of oncogene amplification in human primary breast tumours enabled us to pinpoint three oncogene combinations with different consequences in terms of recurrence and survival. Experimental models would be valuable to understand the molecular mechanisms underlying the described effects.

Rt~SUMI~ PARADIOME POUR LA COMPLt~MENTATION DES ONCOGI~NES DANS LE CANCER MAMMAIRE HUMAIN

Nous avons montr6 que la co-amplification des oncog~nes c-myc et erbB-2 d'une part et c-myc et int-2 d'autre part intervenait de fa~on statistiquement significative dans un 6chantillon de 170 carcinomcs mammaires infiltrants. Un cffet-dose &ant

ONCOGENE COMPLEMENTA TION IN BREAST CANCER

589

habituellement 6voqu6 pour rendre compte de l'activit6 biologique des produits des g~nes amplifi6s, nous avons recherch6 les effets synergiques de la pr6sence ou de l'absence d'amplification sur le devenir clinique des patientes en termes d'6volution de la maladie canc6reuse (diss6mination m6tastatique) et de survie globale. L'amplification de c-myc a ~t~ associ6e ~ une reprise ~volutive pr~coce de la maladie dans les 30 premiers mois alors que des copies surnum6raires des g~nes erbB-2 et int-2 s'accompagnent d'un allongement de l'intervalle libre, les 6v6nements intervenant de fa9on significative entre 30 et 50 mois. L'dvaluation du statut (amplifi6 ou non) intratumoral de ces oncog~nes pris deux ~ deux, a mis en 6vidence l'existence de ph6nom~nes de compl6mentation: (1) positive, la co-amplification de c-myc et erbB-2 6tant associ6e ~ une diminution de la survie globale ; (2) n6gative, l'absence d'alt6rations de la zone int-2 associ6e ~t l'amplification de c-myc r6sultant dans le d61ai de rechutes le plus court; et (3) s6quentielle, l'amplification de erbB-2 ou int-2 en l'absence d'alt6rations de c-myc r6sultant dans un taux de rechutes ~ 50 mois, similaire h celui associ6 ~t l'amplification de c-rnyc en l'absence d'alt6rations de erbB-2 ou int-2, les effets observ6s 6tant toutefois ddcal6s dans le temps et les 6v6nements corr616s h l'amplification de c-myc survenant plus pr6cocement. L'ensemble de ces observations rend compte de certains aspects li6s ~t l'histoire naturelle de la maladie. MOTS-CLI~S: Oncog~ne, Amplification, Compl6mentation, Tumeur; Paradigme, Cancer mammaire, Evolution.

ACKNOWLEDGEMENTS We thank N. Dussault for superior technical assistance. We also thank S. Romain for collecting clinical data and performing computer analysis. We acknowledge H. Benali for valuable comments on statistical analysis. We are indebted to N. Pourreau-Schneider for English review of the manuscript. This work was supported by grants from the Minist6re de la Recherche et de la Technologie and the Ligue Nationale Fran~aise Contre le Cancer.

REFERENCES ALITALO, K., SCHWAB, M., LINN, C.C., VARMUS, H.E. & BISHOP, J.M. (1983), Homogeneous staining chromosomal regions contain amplified copies of an abundantly expressed cellular oncogene (c-myc) in malignant neuroendocrine ceils from a human colon carcinoma. Proc. nat. Acad. Sci. (Wash.), 80, 1707-1711. ANDRES,A.C., SCHOENENBERGER,C.A., GRONER,B., HENNIGHAUSEN,L., LEMEUR,M. GERLINOER, P. (1987), Has-ras oncogene expression directed by a milk protein gene promoter. Tissue specificity, hormonal regulation and tumor induction in transgenic mice. Proc. nat. Acad. Sci. (Wash.), 84, 1299-1303. BLOOM, H.J.G. & RICHARDSON,W.W. (1957), Histological grading and prognosis in breast cancer. Brit. J. Cancer, 11, 359-377. BRUNET, J.F., ROUX-DOSSETO,M., DENIZOT, F., MATTEI, M.G., CLARK, W., HAQQI, T.M., FERRIER, P., NABHOLZ,M., SCHMITT-VERHULST,A.M., LUCtA-~I,M.F. 8£ GOLSTEIN, P. (1986), The inducible cytotoxic T-lymphocyte-associated gene transcript CTLA-1 sequence and gene localization to mouse chromosome 14. Nature (Lond.) 322, 268-271. CAVANEE, W.K. & HANSEN, M.F. (1986), Molecular genetics of human familial cancer, in "Molecular Biology of H o m o Sapiens". CoMHarb. Spr. Syrnp. Quant. Biol. 51, 829-835.

590

M. R O U X - D O S S E T O A N D P . M . M A R T I N

CASEY, G., SMITH, R., McGILLIVRAY,D., PETERS, G.R. & DICKSON, E. (1986), Characterization and chromosome assignement of int-2, a potential protooncogene. MoL Cell. Biol., 6, 502-510. COMPERE, S.J., BALDACCI,P., SHARPE, A.H., THOMPSON,T., LAND,H. & JAENISCH, R. (1989), The ras and myc oncogenes cooperate in tumor induction in many tissues when introduced into midgestation mouse embryos by retroviral vectors. Proc. nat. Acad. Sci. (Wash.), 86, 2224-2228. Cox, D.R. (1972), Regression models and life tables. J. roy. Stat. Soc., 34, 187-220.

ESCOT, C., THEILLET, C., LIDEREAU,R., SPYRATOS,F., CHAMPENE,M.H., GEST, J. 8Z CALLAHAN,R. (1986), Genetic alteration of the c-myc protooncogene (MYC) in human primary breast carcinomas. Proc. nat. Acad. Sci. (Wash.),

83, 4834-4839.

FINLAY,K.A., HtNDES, P.W. & LEVtNE, A.J. (1989), P53 protooncogene can act as a suppressor of transformation. Cell, 57, 1083-1093. FEINBERG, A.P. & VOGELSTEIN, B. (1983), A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity. Analyt. Biochem., 132, 6-13. HOUWELING,VAN DENELSEN,P.J. & Van der EB, A. (1980), Partial transformation of primary rat cells by the leftmost fragment of adenovirus 5 DNA. Virology, 105, 537-550. ISOBE, M., EMANUEL,B.S., GIVOL,D., OPEN, M. & CROCE, C.M. (1986), Location of gene for human p53 tumor antigen to band 17p13. Nature (Lond.), 320, 84-85. KAPLAN,E.L. & MEIER, P. (1958), Nonparametric estimation for incomplete observations. J. Amer. Stat. Assoc., 53, 457-481. KRAINER,A.R., MANIATIS,T., RUSKIN, B. & GREEN,M. (1984), Normal and mutant [~-globin gene pre-mRNAs are faithfully and efficiently spliced in vitro. Cell, 36, 993-1005. LAND, H., PARADA,L.F. & WEINBERG,R.A. (1983), Tumorigenic conversion of primary embryo fibroblasts require at least two cooperating oncogenes. Nature (Lond.), 304, 596-602. LAND, H., PARADA,L.F. & WEINBERG,R.A. (1984), Cellular oncogenes and multistep carcinogenesis. Science, 222, 771-778. LAW, M.I., KAO, F.T., WEI, Q., HARTZ, J.A., GREENE, G.L., ZARUCKISCHLZ,Z.,

CONNEELY,O.M., JONES, C., PUCK, T.T., O'MALLEY, B.W. • HORWITZ,

K.B. (1987), The progesterone receptor maps to human chromosome band 11q13, the site of the mammary oncogene int-2. Proc. nat. Acad. Sci. (Wash.), 84, 2877-2881. LAWLESS, J.F. (1982), Statistical models and methods for lifetimes data (pp. 278-279). John Wiley & Sons, New York. LIDEREAU, R., CALLAHAN, N., DICKSON, C., PETERS, G., ESCOT, C. & ALl, I.U. (1988), Amplification of int-2 gene in primary human breast cancer. Oncogene Res., 2, 285-289. MOOLGAVKAR,S.H., DAY,N.E. & STEVENS, R.G. (1986), Two-stage model for carcinogenesis : epidemiology of breast cancer in females. J. nat. Cancer Inst., 65, 559-569. MATTEI, M.G., PETKOVITCH,M., MATTEI, J.F., BRAND, N. & CHAMBON,P. (1988), Mapping of the human retinoic acid receptor to the q21 band of the chromosome 17. Hum. Genet., 80, 186-188. PETROWTCH, M., BRA~D, N.J., KRUST,A. & CHAMBON,P. (1987), A human retinoic acid receptor which belongs to the family of nuclear receptors. Nature (Lond.), 330, 444-450. PETO, R., PIKE, M.C., ARMITAGE, P., BPESLOW, N.E., Cox, D., HOWARD,S.V., MANTEL, N., McPHERSON, K., PETO, J. & SMITH, P.G. (1977), Design analysis of randomized clinical trials requiring prolonged observations of each patients. - - II. Analysis and examples. Brit. J. Cancer, 35, 1-39.

ONCOGENE COMPLEMENTATION IN BREAST CANCER

591

RASSOULZADEGAN,M., COWIE, A., CARR, A., GLAICHENHAUSS,N., KAMEN,R. & CuzIN, F. (1982), The role of individual polyoma virus early proteins in oncogenic transformation. Nature (Lond.), 300, 713-715. RASSOULZADEGAN, M., NAGHASHFAR, Z., COWIE, A., CARR, A., GRISONI, M., KAMEN,R. • CUZIN, F. (1983), Expression of the large T protein of polyoma virus promotes the establishment in culture of "normal" rodent cell lines. Proc. nat. Acad. Sci. (Wash.), 80, 4354-4358. ROUX-DOSSETO,M., AUFFRAY,C., LILLIE,J.W., Boss, J.M., COHEN,D., DEMARS,R., MAWAS,C., SEIDMAN,J.G. & STROM1NGER,J.L. (1983), Genetic mapping of a human class II antigen [3-chain cDNA clone to the SB region of the HLA complex. Proc. nat. Acad. Sci. (Wash.), 80, 6036-6040. ROUX-DOSSETO,M., ROMAIN,S., DUSSAULT,N. & MARTIN,P.M. (1989), Correlation of erbB-2 gene amplification with low levels of estrogen and/or progesterone receptors in primary breast cancer: Do erbB-2 products delineate hormoneindependent tumors? in "Colloque INSERM-J. Libbey" (John Libbey Eurotext) (in press). RULEY, H.E. (1983), Adenovirus early region 1A enables viral and cellular transforming genes to transform primary cells in culture. Nature (Lond.), 304, 602-606. SCHECHTER, A.L., HUNG, M.C., VAIDYANATHAN,L., WEINBERG, R.A., YANG-FENG, T.L., FRANCKE,U., ULLR1CH,A. & COUSSENS,L. (1985), The neu gene: an erbB-2-homologous gene distinct from and unlinked to the gene encoding the EGF receptor. Science, 229, 976-977. SCHOENENBERGER,C.A., ANDRES,A.C., GRONER,B., Van der VALK,M., LEMEUR,M. t~ GERLINGER, P. (1988), Targeted c-myc gene expression in mammary glands of transgenic mice induces mammary tumors with constitutive milk protein gene transcription. EMBO J., 7, 169-175. SEMBA,K., KAMATA,N., TOYOSHIMA,K. & YAMAMOTO,T. (1985), A v-erbB-related proto-oncogene, c-erbB-2, is distinct from c-erbB-1/epidermal growth factorreceptor gene and is amplified in human salivary gland adenocarcinoma. Proc. nat. Acad. Sci. (Wash.), 82, 6497-6501. SINN, E., MULLER, W., PATTENGALE,P., TEPLER, I., WALLACE, R. & LEDER, P. (1987), Co-expression of MMTV/v-Ha-ras and MMTV/c-myc genes in transgenic mice: synergistic action of oncogenes in vitro. Cell, 19, 181-188. SLAMON,D.J., GODOPHIN,W., JONES, L.A., HOLT, J.A., WONG, S.G., KErrH, D.E., LEVIN, W.J., STUART,S.G., UDOVE,J., ULLRICH,A. & PRESS, M.F. (1989), Studies of the HER/neu proto-oncogene in human breast and ovarian cancer. Science, 244, 707-712. VAN DEL ELSEN, P.J., DE PATER, S., HOUWELING,A., VAN DER VEER, J. & VAN DER EB, A. (1982), The relationship between region E1A and E1B of human adenovirus in cell transformation. Gene, 18, 175-181. WATSON, R., OSKARSSON,M. t~ VANDERWOUDE,G.F. (1982), Human DNA sequence homologous to the transforming gene (Mos) of Moloney murine sarcoma virus. Proc. nat. Acad. Sci. (Wash.), 79, 4078-4082. YAMAMOTO,T., IKAWA, S., ALIYAMA,T., SEMBA, K., NOMURA,N., MIYAJIMA,N., SAITO, T. t~ TOYOSHIMA,K. (1986), Similarity of protein encoded by the human c-erbB-2 gene to epidermal growth factor receptor. Nature (Lond.), 319, 230-234. YOSHIDA,M.C., WADA,M., SATHO,H., YOSHIDA,T., SAKAMOTO,H., MIYAGAWA,K., YOKOTA,T., KODA,T., KAKINUMA,M., SUOIMURA,T. & TERAOE,M. (1988), Human HST1 (HSTF1) gene maps to chromosome band llq13 and coamplifies with INT2 gene in human cancer. Proc. nat. Acad. Sci. (Wash.), 85, 4861-4864. ZAKUT-HOURI,R., BIENZ-TADMOR,B., GIVOL, D. & OREN, M. (1985), Human p53 cellular tumor antigen cDNA sequence and expression in COS cells. EMBO J., 4, 1251-1255.