Breast CancerResearch and Treatment19: 85-93, 199t. © 1991 KluwerAcademic Publishers. Printedin the Netherlands. Report

Prognostic value of total cathepsin D in breast tumors A p o s s i b l e role in selection o f c h e m o r e s i s t a n t p a t i e n t s

Mo~se Namer, Alain Ramaioli, Xavier Fontana, Marie-Christine Etienne, Michel Hdry, Alain Jourlait, G6rard Milano, Marc Frenay, Eric Francois, Fran~oise Lapalus Centre Antoine-Lacassagne, 36 Voie Romaine, 06054 Nice Cddex, France

Key words: adjuvant chemotherapy, cathepsin D, drug resistance, prognostic factors

Summary Evaluation of prognostic factors for breast cancers is important for therapeutic decisions both at the time of surgery and during postoperative surveillance. In 1979, H. Rochefort described an induced protein with a molecular weight of 52,000 Daltons identified as procathepsin D. Total cathepsin D (TCD) (52K + 48K + 34K), expressed in pmol/mg protein, can be measured by an immunoradiometric method commercialized by Cis-Biointernational. Total cathepsin D was assayed in 413 breast cancer tumors from patients who underwent surgery between January 1, 1978, and December 31, 1985. Using a cut-off of 35 pmol/mg protein, patients with an elevated level had a significantly poorer survival than those with a low level (p = 0.03). This difference was not found for node-negative patients but was very significant for node-positive patients (p < 0.008). The survival of node-positive patients with a low total cathepsin level was not statistically different from that of node-negative patients. Analysis of the N+ subgroup of patients who did not receive adjuvant chemotherapy revealed that TCD no longer had any prognostic value, whereas it was still important for the N + subgroup who received an adjuvant treatment. Cox multivariate analysis of prognostic value for survival placed total cathepsin D in third position, after nodal invasion and progesterone receptor status, for the entire population, and in first position before progesterone receptor status for the node-positive population. The association of a low cathepsin level and positive progesterone receptors characterized the subgroup of patients with the longest survival. TCD levels played the same role for prediction of the outcome of metastasis. In conclusion, assays of total cathepsin D in tumor cytosots provide valuable prognostic information for breast cancer patients with nodal involvement at the time of diagnosis.

Introduction When a patient with breast cancer presents at an operable stage, the oncologist's first concern is to determine whether or not the disease is likely to relapse after exeresis. Evaluation of various prognostic factors is one means of predicting the natural history of the tumor. Certain widely used criteria have already proven their efficacy: tumor size [1],

involvement of the axillary lymph nodes [2], the degree of histologic differentiation [3], and the estrogen (ER) and progesterone (PR) receptor status [4]. Other less commonly employed factors include measurement of the proliferative fraction (by assaying the tritiated thymidine index [5] or flow cytometry [6]), assays of growth factors such as EGF receptors [7], and the search for oncogene amplification [8] or expression.

Addressfor offprints: M. Namer, Centre Antoine-Lacassagne, 36 Voie Romaine, 06054Nice C6dex, France

86

M Namer et al.

Nevertheless, no parameter is of absolute reliability. While node-negative cancers have a good prognosis, some 20-30% of them do recur; likewise, node-positive cancers have a less favorable prognosis, but 30-40% of these patients have not relapsed at 10 years. The same situation applies to the degree of histologic differentiation and hormone receptor status. New parameters are thus still required to improve prognostic evaluations and guide therapeutic decisions. In 1979, H. Rochefort [9] described an induced cytosolic protein with a molecular weight of 52,000 Daltons (52 KD) which is stimulated by estradiot and inhibited by antiestrogens in E R + breast cancer cells. This protein was identified as a precursor of a lysosomal enzyme, cathepsin D. Procathepsin D is secreted in excess amount by mammary carcinoma cells as compared with normal mammary cells. This 52 KD precursor is successively processed into a 48 KD (intermediate form) and then into a 34 KD (mature form) plus a 14 KD (short form). Recently, an immunoradiometric assay allowing the quantification of total cathepsin D (TCD), i.e. 5 2 K D + 4 8 K D + 34KD, was developed and commercialized by CIS Biointernational, Gif-surYvette, France. This assay uses two different monoclonal antibodies (D7E3 and MIG8) which recognize two different epitopes of the mature 34KD form [11]. In order to test the prognostic significance of this new factor, TCD assays were performed in 413 stored breast tumor cytosols.

Material and methods

Between January 1, 1978, and December 31, 1985, 862 patients underwent surgery at the AntoineLacassagne Cancer Center for unilateral invasive breast cancer without metastatic spread or any other associated malignancy (whole population). Almost all breast tumor specimens were originally submitted in our laboratory for steroid receptor analysis. Cytosols were prepared in Tris EDTA molybdate buffer, then frozen in liquid nitrogen.

TCD was assayed with the CIS kit (ELSA-cath D Kit, CIS Biointernational) for the 413 patients whose excess cytosots were still available (population 1). Absence of tumor specimens prevented TCD assays for the remaining 449 patients (population 2). Table 1 shows the distribution of the conventional prognostic factors for these two populations. Tumor size (T) was coded using the TNM classification. Nodal status (N) represented the presence or absence of histologically confirmed nodal involvement. Histologic differentiation was coded using the classification of Scarff, Bloom, and Richardson (SBR) [12]. Estrogen and progesterone receptors (ER, PR) were assayed using the dextran-coated charcoal method (DCC) recommended by the EORTC [13]. Cytosol proteins were assayed according to the technique of Bradford [14]. The cut-off value for positivity was set at 10 fm/mg protein for ER and at 15 fm/mg protein for PR. Histologic differentiation could not be performed in 68 patients of the population t and 74 patients of population 2. Similarly, 23 and 26 patients missed their ER and PR evaluations. Median follow-up was 68 months.

Cathepsin D assays

In the TCD sandwich technique assay used, two monoclonal antibodies were prepared against two different antigenic sites of the cathepsin D molecule. The first antibody is coated on the solid phase (ELSA tube) and the second is labelled with 125Ias a tracer. The amount of radioactivity bound to the tube is proportional to the amount of cathepsin D in the test sample. The cytosols were diluted to 1/40 and 1/80, and each dilution was assayed in duplicate. To tubes containing 300/xl anticathepsin D monoclonal antibody labelled with 125Iwere added 50/zl of diluted cytosol. After incubation for 3 h under agitation at ambient temperature, the tubes were washed and the radioactivity was counted. For each cytosol, the mean results were multiplied by the dilution factor and expressed in pmol/mg protein to compare the

Cathepsin D in breast tumors TCD level with the cytosol protein concentration. Each assay was performed in duplicate. Results were expressed in pmol/mg of cytosolic protein.

87

'30 " 9080*0,50 -

Statistical methods

5S40-

The distributions of prognostic factors were compared by the chi-square method. The median values of TCD were compared with the Mann and Whitney test. Survival curves were established by the Kaplan Meier method, and were compared with the Log-rank test [15]. Comparison of the different prognostic factors was performed by Cox univariate and mutivariate analysis [16]. BMDP software was used for all statistical calculations [17].

Pop1: patients with TCO assay: 413 Pop2: palienls without TCD assay: 449

• 20

Log rank test: p=O.09

1 ~1

o

12

24

36

48

I

]

t

00

72

84

.....

I

g6

l

I

108

t20

Month=

Fig. 1. Comparison of the survival of populations i and 2.

Results

1. Comparison of populations 1 and 2

Table 1. C o m p a r i s o n of the distribution of the conventional prognostic factors in the population submitted to T C D assay and the remaining population who had no T C D assay Parameters

Patients without T C D assay. Population 2 (N = 449)

Patients with T C D assay. Population 1 (N = 413)

Number

%

Number

%

T0--1 T2 T3-4

153 250 46

34 56 10

90 267 56

22 65 13

0,0003

NN+ N unknown

284 165 0

63 37 -

246 166 1

60 40 -

0.28

SBR SBR SBR SBR

135 190 50 74

30 42 11 17

121 178 46 68

29 43 11 17

ER+ ERER unknown

320 106 23

71 24 5

312 100 1

76 24 -

0.83

PR+ PRPR unknown

286 137 26

64 30 6

297 115 1

72 28 -

0.15

I II III unknown

* Chi square test.

p value*

Comparison of the distribution of conventional prognostic factors in the two populations shows that tumors for which TCD assays were performed were bigger than the tumors of the other patients (Table 1). This is logical since small tumors are less likely to yield sufficient tumor cells for retrospective studies. Nevertheless, no significant difference was seen in patient survival (Fig. 1).

2. TCD levels

0.96

Figure 2 illustrates the asymmetrical distribution of TCD levels and Table 2 the median TCD concentration in various situations. The median TCD value was significantly higher in N + patients than in the N - population (p -- 0.04), but it was not significantly different in patients who developed metastases as compared with those who had a favorable outcome. Similarly, there was no difference between premenopausal and postmenopausal patients.

3. Selection of TCD cut-off level For analysis of survival of population 1, determina-

88

M Namer et al. 1O0

80

60 t,l

"6 ~z

40

20

__90

Cathepsin concentration (pmol/mg protein)

Fig. 2, Distribution of TCD values.

tion of the most important discriminant value around the median point, between 15 and 70pmol/ mg protein, led to definition of 35 pmol/mg protein as the optimal cut-off. Identification of the best cut-off value for patients as a function of nodal status revealed the same value (35 pmol/mg prot) for N+ patients (Fig. 3); no statistically discriminant value was found .for N-patients. p~aiue

4. Value of TCD assays/Or predicting survival Comparison of the survival of patients with a TCD level > 35 pmol/mg protein (209 patients) with that of patients who had -< 35 pmol/mg protein (204 patients) revealed a benefit in favor of the latter group (p = 0.03) (Fig. 4). Cox proportional hazard analysis of conventional prognostic parameters including the TCD value, revealed that nodal invasion was the predominant factor, followed imTable 2. Median total cathepsin D concentration

01

Number patients

Median (pmol/mg protein)

p value*

Entire population

413

35.3

NN+

246 166

33.7 38.0

M0 M1

303 110

34.7 37.3

0.24

Premonopausal Postmenopausal

t00 313

34.4 35.4

0.93

Fig. 3. Maximal likelihood determination of the cut-off of TCD for predicting survival of N+ breast cancer patients. P values obtained for each cut-off value plotted against the value itself. Statistical significance is indicated by the horizontal line at the 0.05 level.

* Mann and Whitney test. M0: Patients without metastases during follow-up. MI: Patients with metastases during follow-up.

0 05

0.01 TCD concentration o1o

15

2o

2s

30

35

40

45

50

55

60

65

~0

0.04

pmoflrng prot.

Cathepsin D in breast tumors 100--

89

100 '

=

90-

"

90 -

"

80-

80"

70-

70-

60-

60"

50,

50' 4o ¸

40

30

B

TCD s35 204 p~s TCD>35:209 pls

•

u Log

20

20-

rank t e s t

N- and N and N+ and N+ and

TCD TOO TCD TCD

!;35 ~32 pls >35: 11,~ p~s 43S: 72 pts > 3 5 : 9 4 pts

p35:39 pts

3020-

2O

Leg rank test: p= 0.25

Log r a n k test: p= 0.01 I0

0

12

24

36

48

60

72

84

96

~2

24

36

48

6(}

72

84

Months

96 Months

Fig. 6. Survival of N+ patients who received adjuvant chemotherapy (a) and who did not receive adjuvant chemotherapy (b).

were not correlated with any difference in survival, regardless of the TCD concentration. By contrast, P R + patients with a low TCD value had a better survival than those with a higher TCD level (p = 0.006). Survival of P R + patients with a high TCD level was as poor as the survival of P R - patients (Fig. 7).

5. Value of TCD for predicting metastasis-free survival The difference noted for the influence of TCD levels in the survival of node-negative and nodepositive patients was also seen in the onset of metastasis (Fig. 8). The four metastasis-free survival curves are globally different (p < 0.00001). Nevertheless, while the metastasis-free survival of nodepositive patients was favorably influenced by a low TCD level (Log-rank test: p < 0.02), the onset of metastasis was not influenced by the TCD level in node-negative patients. Cox proportional hazards analysis for the entire population identified nodal involvement and progesterone receptor status as prominent prognostic factors (Table 5); for the N+ patient subgroup, the TCD level was the sole parameter selected (p = 0.03).

TCD appears to be an interesting prognostic factor for predicting the outcome of patients who undergo surgery. TCD was present in all 413 patients (246 nodenegative, 166 node-positive). The median level was 35.3 pmol/mg protein and the best cut-off defining two different populations in term of survival was 35pmol/mg protein. Definition of an elevated TCD level allowed identification of a subgroup Table 4. N+ population. Comparison of the distribution of the conventional prognostic factors in the treated population and the untreated population Parameters

Treated

Untreated

No

%

No

%

95

100

71

100

18 53 24

19 56 25

14 48 9

20 67 13

19 47 17 12

20 49 18 13

22 35 4 10

31 49 06 14

ER+ ER-

68 26

72 28

61 10

86 14

PR+ PR-

65 29

69 31

59 12

83 17

Total T0-1 T'2 T3-4 SBR SBR SBR SBR

I II III unknown

Discussion

Mean age (yr)

A protease that can be assayed in breast tumors,

* Chi square test.

54 ± 12

66 :L 11

p value*

0.12

0.09

0.03

91

Cathepsin D in breast tumors 100 -

80-

7

70-

60-

PR* and TOD 435:134 PR+ and TCD >35:163 PR- ,and TCO " 4 3 5 : 6 9 PR- and TCD > 3 5 : 4 6

•

m e

50"

pIs pTs p;s p[s

40.

=1 * Ii e

30

30

2O

20

Log rank test: p=0.0007

N- a~o TCD N and TCD N÷ a~-d TCD N* ar'd TCD

' 3 5 !32 >35; ; / 4 -35:72 >35 94

p~a 9t$ pts pts

Log rank test: p