Response of Prostate Cancer to Anti-Her-2/neu Antibody in Androgen-dependent and -independent Human Xenograft Models¹

Her-2/neu is overexpressed in 25–30% of human breast and ovarian cancers and predicts for a poor prognosis in breast cancer patients (1, 2, 3). Antibodies to the extracellular domain of Her-2/neu (gp185) can inhibit the growth of tumor xenografts and transformed cells that express high levels of this tyrosine kinase receptor (4, 5, 6, 7). A recombinant, humanized mAb³ anti-P185^Her-2/neu (Herceptin) has a cytostatic growth-inhibitory effect on breast cancer cells overexpressing Her-2/neu (8) and has been demonstrated to enhance the antitumor activity of paclitaxel and doxorubicin against Her-2/neu-overexpressing human breast cancer xenografts (9, 10). This antibody has clinical activity in breast cancer patients with metastatic disease overexpressing Her-2/neu (11) and has demonstrated clinical activity when combined with paclitaxel in patients with breast cancer. Because Her-2/neu may be overexpressed in prostate cancer (12, 13, 14, 15), we examined the effect of Herceptin monotherapy in androgen-dependent (CWR22 and LNCaP) and androgen-independent (CWR22 derivatives) prostate cancer xenograft models (16, 17, 18) and the combination of Herceptin with paclitaxel. Our results indicate that clinical trials of Herceptin in human prostate cancer are warranted.

Four to six-week-old nude athymic BALB/c male and female mice were obtained from the National Cancer Institute-Frederick Cancer Research & Development Center and maintained in pressurized ventilated caging at the Sloan-Kettering Institute. Male animals were inoculated s.c. with 1 × 10⁶ LNCaP cells or minced tumor tissue from the androgen-dependent CWR22 cell line, and females received the androgen-independent sublines CWR22R, WR22SA1, CWRSA4, or CWRSA6, which were obtained by selecting tumors for regrowth and increased serum PSA after androgen withdrawal. All cell lines were injected together with reconstituted basement membrane (Matrigel; Collaborative Research, Bedford, MA), as described previously (16, 17, 19). To maintain serum testosterone levels, male mice were implanted with 12.5-mg sustained release testosterone pellets (Innovative Research of America, Sarasota, FL) s.c. before receiving the tumor cell inoculation. Treatments consisted of a twice weekly i.p. injection of 20 mg/kg Herceptin (recombinant, humanized mAb Her2; Genentech, Inc., South San Francisco, CA) in PBS for no less than 3 weeks and/or paclitaxel (Taxol; Bristol Myers-Squibb Co., Princeton, NJ) low dose (6.25 mg/kg s.c., five times/week × 3 weeks) or high dose (12.5 mg/kg s.c., five times/week × 2 weeks) in sterile saline. Control mice were given vehicle alone. Tumors were measured every 3–4 days with vernier calipers, and tumor volumes were calculated by the formula π/6 × larger diameter × (smaller diameter) (Ref. 2). Animals with palpably established tumors of at least 65 mm³ in volume were designated to treatment groups.

Xenografts were assayed for Her-2/neu expression by immunohistochemistry using the DAKO Her-2 kits (HercepTest; DAKO Corp., Carpinteria, CA). The samples were scored blindly by comparison with standard controls in the DAKO kit standards and scored as follows: (a) 0, no staining or membrane staining in <10% of the tumor cells; (b) 1⁺, faint membrane staining in >10% of the tumor cells; (c) 2⁺, weak to moderate complete membrane stain in >10% of cells; or (d) 3⁺, moderate to strong complete membrane staining in >10% of cells). A score of 0 or 1⁺ was considered negative for Her-2/neu overexpression, whereas 2⁺ or 3⁺ indicated Her-2/neu overexpression. FISH analysis was done using the Oncor kits (Inform Her-2/neu gene detection system; Oncor, Gaithersburg, MD). A minimum of 100 tumor cells in each tumor were evaluated for nuclear Her-2/neu gene copy number (13).

Blood samples (∼50 μl) from male mice collected in microtainer serum separator tubes (Becton Dickinson, Franklin Lakes, NJ) by superficial incision of the dorsal tail vein were taken before therapy and on days 9 and 21 of treatment. PSA values were then determined from the serum using the Tandem-R PSA immunoradiometric assay (Hybritech, San Diego, CA).

Pairwise differences between the tumor volumes of the treatment groups were compared over time using a permutation test. The null hypothesis for this test is that treatment has no differential effect on the tumor volumes over time. The statistic used to test the hypothesis was the sum of the squared differences between mean tumor volume summed over all time points.

SS_Dev was used to capture the average differences between treatment groups at each time point. This statistic reflects the amount by which the trajectories of average tumor volume of the two treatment groups are different.

The Her-2 expression patterns of the androgen-dependent and androgen-independent prostate xenografts were examined by immunohistochemistry and FISH. The parental androgen-dependent CWR22 tumors demonstrated 2⁺ Her-2/neu staining, and the LNCaP tumors demonstrated 3⁺ Her-2/neu staining (Fig. 1, A and B). The androgen-independent sublines of CWR22 demonstrate 2⁺ (CWRSA1), 3⁺ (CWRSA4), 2⁺ (CWRSA6), and 1⁺ (CWR22R) staining for Her-2/neu (Fig. 1, C–F). All tumors had a 2–4 Her-2/neu gene copy (normal range) number by FISH (data not shown).

We evaluated the efficacy of Herceptin in well-established androgen-dependent and androgen-independent prostate cancer xenografts. The CWR22, LNCaP, CWR22R, and CWRSA6 models were used for these experiments because they provided reproducible growth curves. Herceptin (20 mg/kg twice weekly) had no effect on the tumor growth curve in any of the androgen-independent tumors when compared to control (CWR22R, P = 0.60, n = 10, Fig. 2,A; CWRSA6, P = 0.63, n = 10, Fig. 2,B). The murine anti-Her-2 mAb, 4D5, also had no effect on tumor growth in the CWR22R androgen-independent cell line (P = 0.21, n = 10; data not shown). In contrast, Herceptin did show significant growth inhibition of both of the androgen-dependent xenograft models, CWR22 (68% growth inhibition, P = 0.03, n = 12, Fig. 2,C) and LNCaP (89% growth inhibition, P = 0.002, n = 12, Fig. 2 D). The histology of the tumors in each group was examined by light microscopy, and no visible difference in necrosis was seen between the groups (data not shown).

Graded doses of paclitaxel were administered to the xenograft-bearing animals to determine a dosage of paclitaxel that inhibited growth but did not cause complete regression of the tumors (data not shown). Paclitaxel at a dosage of 6.25 mg s.c. five times/week given to animals with well-established tumors caused growth inhibition as compared to control at the time of the sacrifice (CWR22, 76% growth inhibition, P = 0.02, Fig. 2,E; LNCaP, 84% growth inhibition, P = 0.002, Fig. 2,F; CWR22R, 76% growth inhibition, P = 0.02, Fig. 2,G; CWRSA6, 51% growth inhibition, P = 0.02, Fig. 2,H). When paclitaxel and Herceptin were coadministered to animals, there was a marked reduction in tumor volume versus control for both androgen-dependent and androgen-independent tumors (CWR22, 98% growth inhibition, P = 0.01, Fig. 2,E; CWR22R, 92% growth inhibition, P = 0.01, Fig. 2,G; LNCaP, 94% growth inhibition, P = 0.006, Fig. 2,F; CWRSA6, 77% growth inhibition, P = 0.01, Fig. 2,H). Increased growth inhibition was observed with the combination of Herceptin and paclitaxel as compared to each agent alone at the end of the treatment period in the animals with androgen-dependent xenografts (Fig. 2, E–H), the CWR22 group [mean tumor volumes (n = 6 in each group): paclitaxel, 408 mm³; Herceptin, 520 mm³; paclitaxel and Herceptin, 76 mm³; paclitaxel versus paclitaxel and Herceptin, P = 0.03] and the LNCaP group [mean tumor volumes (n = 6 in each group): paclitaxel, 233 mm³; Herceptin, 163 mm³; paclitaxel and Herceptin, 82 mm³; paclitaxel versus paclitaxel and Herceptin, P = 0.03]. In addition, there was increased growth inhibition with the combination of Herceptin and paclitaxel versus each agent alone at the end of the treatment period in the animals with androgen-independent xenografts (Fig. 2, E–H), the CWRSA6 group [mean tumor volumes (n = 5 in each group): paclitaxel, 1496 mm³; Herceptin, 2941 mm³; paclitaxel, and Herceptin, 687 mm³; paclitaxel versus paclitaxel and Herceptin, P = 0.001) and the CWR22R group [mean tumor volumes (n = 5 in each group): paclitaxel, 1273 mm³; Herceptin, 3811 mm³; paclitaxel and Herceptin, 592 mm³; paclitaxel versus paclitaxel and Herceptin, P = 0.095].

The lack of clinical activity of Herceptin in the androgen-independent models could indicate a requirement for androgen to be present for the Herceptin antitumor response to be manifested. Therefore, androgen pellets were added to the androgen-independent models (CWR22R and CWRSA6). The response to Herceptin and Herceptin with paclitaxel was no different from that in the absence of androgen (data not shown).

The PSA index (ng PSA/ml serum/mm³ tumor) was measured in the androgen-dependent animals at the time treatment was initiated (day = 0) and at days 9 and 21 of treatment. There was a significant increase in the PSA index in Herceptin-treated androgen-dependent animals at both time points when compared with pretreatment values, whereas the control animals showed a decrease in the PSA index at either time point when compared to day 0 [LNCaP, control 0.6 relative to pretreatment value; Herceptin group, 2.35 relative to pretreatment value at day 21 (Fig. 3,A); CWR22, control 1.0 relative to pretreatment value; Herceptin group, 18 relative to pretreatment value at day 21 (Fig. 3 B)]. The relative PSA index decreased in the LNCaP untreated group possibly secondary to increased necrosis with increasing tumor size, although this was not detectable by light microscopy. There was also an increase in the PSA index after combination treatment with Herceptin and paclitaxel when compared with pretreatment values (data not shown).

We found that Herceptin alone has antitumor activity in the androgen-dependent human prostate cancer xenograft models studied and has at least an additive effect in combination with paclitaxel in inhibition of tumor growth. Herceptin alone had no demonstrable antitumor effect in the androgen-independent models but clearly enhanced the activity of paclitaxel. The response to Herceptin alone appears to require the presence of androgen and/or that the tumor itself be androgen dependent. The lack of response of the androgen-independent tumors to Herceptin in the presence of androgen indicates that androgen dependence or the requirement for signaling through the androgen receptor is a necessary element for Herceptin response in these prostate tumors.

It has recently been shown that overexpression of Her-2/neu can allow androgen-independent growth in LNCaP xenograft models (20). All of the androgen-independent xenografts used in our studies expressed androgen receptor, as assayed by immunohistochemistry (data not shown). The prostate cancer models used in this study overexpress Her-2/neu on the cell surface. The expression of functional androgen receptor, coupled with the overexpression of Her-2/neu in the parental CWR22 cell line, suggests that Her-2/neu is not expressed because of androgen insensitivity in these tumors. The lack of response to Herceptin in the androgen-independent tumors suggests that the androgen signaling pathway in the androgen-dependent models may be permissive with regard to Herceptin response. The basis for the ability of Herceptin to enhance the paclitaxel antitumor effect in androgen-independent prostate cancer is uncertain. It is possible that Herceptin initiates or impairs cell signaling pathways that allow for increased paclitaxel cell kill. Other mechanisms for this effect may be immunological, or Herceptin may affect the tumor concentration of paclitaxel.

Growth inhibition in the androgen-dependent tumors by Herceptin was accompanied by an increased PSA index when compared with pretreatment values. In contrast, the control-treated animals showed a decrease in the PSA index with time. Her-2/neu overexpression in prostate cancer cells may cause androgen-independent growth by activation of the androgen receptor signaling pathway in a ligand-independent fashion (20). In our model systems, Herceptin binding to Her-2/neu led to increased cellular secretion of PSA in an androgen-independent fashion. This result further supports the notion of cross-talk between the Her-2/neu and androgen receptor signaling pathways. The lack of correspondence between the PSA index and tumor response will complicate clinical trial design, which routinely uses PSA as a surrogate for tumor response. Clinical trials are currently underway examining the role of Herceptin in human prostate cancer using end points other than PSA.

We thank Dr. Victor Reuter and Nita Arroya for FISH analysis of the tumors and Dr. Mark Sliwkowski for gift of Herceptin and helpful advice.