Evaluation of nuclear factorκB and chemokine receptor CXCR4 co-expression in patients with prostate cancer in the Radiation Therapy Oncology Group (RTOG) 8610

Abstract

Objective

• To determine the frequency of nuclear factor κB (NFκB) and the chemokine receptor CXCR4 co-expression in prostate cancer specimens from men with locally advanced disease.

Patients and methods

• Paraffin-embedded samples from patients enrolled on the Radiation Therapy Oncology Group (RTOG) 8610 trial underwent immunohistochemical staining for NFκB and CXCR4.

• The amount of NFκB and CXCR4 was scored by a ‘blinded’ pathologist for the percentage of cells stained (0–100%) and staining intensity (0–3 +).

• Cox proportional hazard models were used for overall survival and disease-free survival to examine if NFκB and/or CXCR4 expression were associated with patient outcomes with and without adjustment for covariates.

Results

• Available material and successful staining allowed NFκB and CXCR4 status to be determined for 55 and 63 patients, respectively.

• Both NFκB and CXCR4 status were available for 51 patients. Of these, 53% were 2/3 + for cytoplasmic NFκB staining and 56% were 2/3 + for CXCR4.

• In all, 18 of the 51 patients were 2/3 + for both NFκB and CXCR4 (P = 0.129). Ten of 11 patients with 3 + NFκB had 2/3 + CXCR4 (P = 0.004).

• In this small study, neither NFκB nor CXCR4 were associated with prostate cancer outcomes.

Conclusion

• High NFκB expression is associated with CXCR4 expression and they are co-expressed in about one third of patients with organ-confined prostate cancer.

• Larger studies to accurately determine the frequency of co-expression and prognostic utility of NFκB and CXCR4 alone and in combination are warranted.

Introduction

Nuclear factor κB (NFκB) is a dimeric transcription factor composed of members of the Rel family [1] and plays important roles in the production of angiogenic, anti-apoptotic and prometastatic factors that are involved in carcinogenesis. The predominant NFκB dimers are the transcriptionally active p65:p50 heterodimer and the less active p50:p50 homodimer [2] and these dimers are bound by inhibitory proteins (IκBs) in the cytoplasm which, after phosphorylation, induce nuclear localization and activation of NFκB. The release of NFκB and subsequent binding to DNA occurs in response to various stimuli, which include cytokines such as TNFα and interleukin 1, growth factors, chemotherapy and radiation [1].

Constitutive activation of NFκB has been shown in an array of malignancies, in particular lymphomas, leukaemias and breast cancer [3,4]. It is increasingly recognised that prostate cancer cells also have constitutive NFκB activity due to increased activity of the IκB kinase complex and an inverse relationship between NFκB activity and androgen receptor status has been reported [5,6]. Furthermore, polymorphisms of the promoter NFκB gene have been reported to be more frequent in patients with prostate cancer compared with controls [7]. Genes activated by NFκB play a central role in many of the hallmarks of cancer including invasion (interleukin 6 and matrix metalloproteinase 9), angiogenesis (interleukin 8, vascular endothelial growth factor) and inhibition of apoptosis: (cIAP 1, c-IAP 2, TRAF-1, TRAF-2, Bfl-1/A1, Bcl-X_L and manganese superoxide dismutase) [8–11]. Emerging preclinical evidence further implicates NFκB in the development of prostate cancer as it has been shown to regulate bcl-2 transcription [12], and inhibition of NFκB results in apoptosis [13] and cell cycle arrest [14].

Moreover, NFκB has been shown to be an important mechanism of hormonal and chemotherapy resistance in prostate cancer cells. Preclinical assessments of plant compounds that inhibit proliferation of prostate cancer cell lines in vitro also implicate modulation of NFκB activity as the mechanism of action [15]. In vitro studies of the NFκB inhibitor, parthenolide, have shown a dose-dependent inhibition of prostate cancer cell proliferation and of interest, resulted in decreased genes associated with carcinogenesis and furthermore, radiosensitized prostate cancer cells with constitutively activated NFκB and improved sensitivity to chemotherapy and hormonal agents [6,15]. In vivo studies have also shown that NFκB activation contributes to development of metastatic prostate cancer and resistance to castration [16,17]. Therapy directed against NFκB in conjunction with other anticancer agents, may therefore be a reasonable target in prostate cancer. It should be noted that bortezomib, a proteasome inhibitor with associated NFκB inhibition properties and major activity in myeloma, did not show any meaningful clinical activity but did have some pharmacodynamic activity in patients with prostate cancer [18,19].

However, the exact targets of NFκB that promote prostate cancer development and/or progression remain largely undefined. The chemokine receptor CXCR4, has been implicated in regulating metastasis of breast, pancreatic and prostate cancer and is under control of NFκB. Predominantly expressed on lymphocytes, CXCR4 is a transmembrane chemokine receptor that activates chemotaxis after binding with CXCL12, the physiological ligand [20]. Historically, it has most been recognised for its role as a co-receptor for HIV but more recently, its roles in growth-regulation, angiogenesis and embryonic development have prompted evaluation into the effect of inhibition on tumour growth and progression. Increased CXCR4 levels have been found in several tumour types, including prostate cancer cells, and may portend a worse prognosis and in particular, correlate with the development of bone metastases [20]. A recent preclinical study of a CXCR4 antagonist reported tumour responses in human prostate cancer cell lines [21] and several other mouse models have reported tumour response to CXCR4 antagonists in cell lines of breast cancer and ovarian cancer [22,23]. Phase I clinical trials of CXCR4 antagonists are now underway and preliminary safety data is awaited.

There is little information on the relationship of CXCR4 and NFκB, and a limited understanding of their influencing roles and mechanisms of action in prostate cancer growth and survival outcomes. The present retrospective cohort study evaluated the relationship of NFκB and CXCR4 in human prostate cancer specimens and correlated them with clinical outcome in patients in the Radiation Therapy Oncology Group (RTOG) 8610 trial.

Patients and methods

RTOG 8610 is a phase III randomized control trial of radiation therapy (RT) alone (RT-alone arm) vs short-term neoadjuvant and concurrent androgen deprivation and RT (STAD + RT arm) in men with locally advanced prostate carcinoma [24]. Paraffin-embedded samples were obtained from a subset of patients enrolled on the RTOG 8610 trial. All patients therefore had organ-confined prostate cancer and were treated with RT, with or without androgen deprivation in the adjuvant setting. Pretreatment characteristics of age, combined Gleason score and clinical stage were collected. Assigned treatment arm was also recorded for each patient with samples subjected to immunohistochemistry. Serial 5 μm-thick sections of radical prostatectomy specimens were used, and the tissue blocks containing the highest Gleason score and the maximum amount of tumour were selected. One representative slide from each case was analysed and the limitation of sample variation was recognized. Human Subjects Approval was obtained at Indiana University.

For the immunohistochemistry, samples were removed from the paraffin by placing them three-times in xylene with subsequent re-hydration through graded ethanol and then immersion in distilled water. Slides were rinsed in Tris buffer saline (TBS) and antigen retrieval performed by using Dako Target Retrieval kit (Dako, Carpinteria, CA, USA) containing a citrate buffer with pH 6.0 for 20 min at ≈95 °C. Dako’s Avidin Biotin blocking system was used for 10 min and the tissue sections were then rinsed with TBS. The nonspecific binding sites were blocked by incubating with Dako’s Protein Block for 20 min.

The antibody used to assess NFκB status was a rabbit polyclonal antibody that identifies the NFκB/p65 (REL A) Ab-1 LabVision Cat#RB-1638. Tissue sections were incubated with the antibody (1: 100 dilution at room temperature for 60 min). After washing with TBS, the secondary antibody, Dako Link (Dako LSAB2 kit) was applied for 20 min and then rinsed with TBS. Additional washing was followed by incubation with streptavadin horseradish peroxidase (HRP; Dako Label, LSAB2 kit) for 20 min. Immunoreactivity was visualized by incubation of sections with diaminobenzidine (DAB) in the presence of H₂O₂. Sections were counterstained with light haemotoxylin and mounted with a coverslip. The antibody employed to assess CXCR4 status was the CXCR4 (Fusin)(C-20) Santa Cruz SC-6190. Again, the prostate cancer specimens were rinsed with TBS and antigen retrieval was achieved with EDTA (pH 8.0) and use of a pressure cooker for 15 min, cooled and rinsed with water and then TBS. The subsequent steps were rinse with 3% H₂O₂ for 15 min, rinse with TBS, expose to the primary antibody at 1: 200 for 60 min, rinse with TBS, apply a secondary antibody (donkey anti-goat 1: 100; Jackson Laboratories) for 30 min, rinse with TBS and expose to streptavidin-HRP Label (Dako LSAB2) for 30 min, rinse with TBS and expose to DAB for 5 min.

The amount of NFκB and CXCR4 was scored by a ‘blinded’ pathologist. The extent and intensity of staining were evaluated in benign epithelium, prostate intraepithelial neoplasia and adenocarcinoma from the same slide for each case. Microscopic fields evaluated and scored were those with the highest degree of immunoreactivity. A numeric intensity score of 0–3 was assigned to each case (0, no staining; 1 +, weak staining; 2 +, moderate staining and 3 +, strong staining). Staining was also dichotomized into negative and positive. Negative was scored if 0 +/1 + and positive was scored if 2 +/3 +.

The endpoints used in the analysis were per protocol; overall survival (OS) and disease-free survival (DFS). The failure event for OS was defined as death from any cause. The OS time was measured from the date of randomization to the date of death or the date of last follow-up (censored). A failure in DFS was defined as death from any cause, local progression, biochemical failure, regional metastasis, or distant metastasis (see the original clinical trials report for details of each outcome [24]). Time to a DFS was measured from the date of randomization to the earliest occurrence of all failure events or the most recent follow-up.

NFκB and CXCR4 are scored by ordinal score (0, 1 +, 2 +, 3 +) and the percentage of cells that stained positively (index percentage). They were also dichotomized as negative (0, 1 +) and positive (2 +, 3 +). The following covariates were considered in the multivariate analyses; age (< 75 years [reference level; RL] vs ≥ 75 years), assigned treatment (RT alone [RL] vs STAD + RT), Gleason Score (2–6 [RL] vs 7–10), and clinical T-stage (T2 [RL] vs T3). Cox proportional hazard models were used for OS and DFS to examine if each NFκB and CXCR4 expression was associated with patient outcomes with and without adjusted for covariates. The pretreatment characteristics and outcomes were compared between the patient groups with and without missing NFκB and CXCR4 values, and between the NFκB and CXCR4 negative and positive groups by chi-square test statistics. Unadjusted and adjusted hazard ratios were calculated for all covariates using either the Cox proportional hazard model with 95% CIs and P-values. The Cochran-Mantel-Haenszel tests were used for the correlation between the dichotomized two expressions and Spearman’s correlation statistics were used for the correlation between the ordinal two expressions. All statistical tests were done at a significance level of 0.05.

Results

Of the 456 patients eligible for enrolment onto RTOG 8610, available material and successful staining permitted NFκB staining in 55 patients and CXCR4 staining in 63 patients. In all, 51 patients had both NFκB and CXCR4 expression determined on their prostatectomy samples and 29 (56%) had positive staining for NFκB (Fig. 1A) and 35 (56%) had positive staining for CXCR4 (Fig. 1B). Of the 51 patients with staining data available for both proteins, 26 (51%) had positive expression for NFκB and 31 (61%) had positive expression for CXCR4. The median follow-up of the 55 patients with NFκB staining was 5.4 years and for the 18 surviving patients was 9.0 years. The median follow-up of the 63 patients with CXCR4 staining was 6.4 years and for the 21 surviving patients also 9.0 years. There were no statistically significant pretreatment characteristic differences between with and without either NFκB or CXCR4 expressions (data not shown). Also, there were no statistically significant pretreatment characteristic differences between negative and positive NFκB groups (Table 1). This was the same for negative and positive CXCR4 groups (Table 1). There was a positive correlation between CXCR4 and cytoplasmic NFκB staining. In all, 18 of 51 patients were positive for both NFκB and CXCR4 (Table 2) and 10 of the 11 patients with 3 + NFκB had positive CXCR4. There was a statistically significant correlation between higher NFκB cytoplasmic expression and higher levels of CXCR4 (Table 2, P = 0.004). However, there was no statistically significant correlation between dichotomized (positive vs negative) NFκB and CXCR4 scores (Table 2, P = 0.21). In all, 18 of 36 patients (69%) with positive NFκB had positive CXCR4 and 12 patients (48%) with negative NFκB had negative CXCR4.

Fig. 1.

NFκB (A) and CXCR4 (B) staining; both ×100.

Table 1.

Pretreatment characteristics of patients by NFκB (n = 55) and CXCR4 intensity score (n = 63)

Characteristics	Intensity score				Negative (0, 1)	Positive (2, 3)	Total	P*
	0	1	2	3
NFκB
N	4	22	17	12	26	29
Age, years								0.47
n/N or n (%):
<75	3/4	18 (82)	14 (82)	7 (58)	21 (81)	21 (72)	42
≥ 75	1/4	4 (18)	3 (18)	5 (42)	5 (19)	8 (28)	13
median (range)	70.5 (51–78)	69.5 (55–79)	73 (58–80)	72.5 (61–81)	70 (51–79)	73 (58–81)
n/N or n (%):
Combined Gleason score (central review)								0.59
2–6	2/2	5 (23)	3 (18)	3 (25)	7 (27)	6 (21)	13
7–10	2/2	17 (77)	14 (82)	9 (75)	19 (73)	23 (79)	42
Clinical stage								0.81
T2	0	7 (32)	5 (29)	2 (17)	7 (27)	7 (25)	14
T3	4/4	15 (68)	12 (71)	10 (83)	19 (73)	22 (76)	41
Assigned treatment								0.52
RT alone	3/4	10 (45)	10 (59)	7 (58)	13 (50)	17 (59)	30
STAD + RT	1/4	12 (55)	7 (41)	5 (42)	13 (50)	12 (41)	25
Total	4 (7)	22 (40)	17 (31)	12 (22)	26 (47)	29 (53)	55
CXCR4
N	8	20	23	12	28	35
Age, years								0.84
n/N or n (%):
<75	4/8	17 (85)	18 (78)	9/12	21 (75)	27 (77)	48
≥ 75	4/8	3 (15)	5 (22)	3/12	7 (25)	8 (23)	15
median (range)	73.5 (62–79)	68.5 (51–77)	70 (55–80)	71 (56–81)	70.5 (51–79)	71 (55–81)
n/N or n (%):
Combined Gleason score (central review)								0.60
2–6	3/8	5 (25)	5 (22)	3/12	8 (29)	8 (23)	16
7–10	5/8	15 (75)	18 (78)	9/12	20 (71)	27 (77)	47
Clinical stage								0.63
T2	1/8	4 (20)	6 (26)	2/12	5 (18)	8 (23)	13
T3	7/8	16 (80)	17 (74)	10/12	23 (82)	27 (77)	50
Assigned treatment								0.74
RT alone	6/8	8 (40)	13 (57)	6/12	14 (50)	19 (54)	33
STAD + RT	2/8	12 (60)	10 (43)	6/12	14 (50)	16 (46)	30
Total	8 (13)	20 (32)	23 (37)	12 (19)	28 (44)	35 (56)	63

^*P-values from Chi-square statistics – Chi-square statistics can not be calculated for the intensity scores (0, 1, 2, 3, 4) because the numbers in some cells are < 5.

Table 2.

CXCR4 intensity by NFκB intensity score and dichotomized CXCR4 intensity score by dichotomized NFκB intensity score (n = 51)

CXCR4	NFκB intensity				Total	NFκB negative (n = 25)	NFκB positive (n = 26)	P
	0	1	2	3
N (%):
CXCR4 intensity:
0	3 (6)	1 (2)	1 (2)	0	5
1	1 (2)	7 (14)	6 (12)	1 (2)	15
2	0	10 (20)	7 (14)	5 (10)	22
3	0	3 (6)	1 (2)	5 (10)	9
Total	4	21	15	11				0.004†
CXCR4 negative (n = 20)						12 (48)	8 (31)	0.21*
CXCR4 positive (n = 31)						13 (52)	18 (69)

^†Spearman correlation statistics;

^*Chi-square statistics.

Table 3 shows the outcomes by NFκB and CXCR4 expression with and without adjusting for other covariates. Dichotomized NFκB intensity score was not a statistically significant prognostic factor for OS or DFS (P = 0.66 and P = 0.77, respectively). The results remained the same when they were adjusted for other covariates (P = 0.34 and P = 0.99, respectively). Dichotomized CXCR4 expression was not a statistically significant prognostic factor for OS or DFS (P = 0.17 and P = 0.77, respectively). There was also no statistically significant difference in OS or DFS adjusted for other covariates (P = 0.19 and P = 0.82, respectively). Figure 2A 3B show the survival curves for each outcome. We examined the outcomes of patients whose NFκB and CXCR4 were both positive vs at least one was negative without and with adjusted for other covariates (Table 4). There were also no statistically significant differences without and with adjusted for other covariates in OS (P = 0.67 and P = 0.77, respectively) and DFS (P = 0.73 and P = 0.85, respectively). However, the numbers of failure event as well as samples were too small to make any statistical conclusion.

Table 3.

Cox proportional hazards regression model by dichotomized NFκB and CXCR4 intensity score

Outcome	Covariate	Comparison	Fail/total	Survival rate at 5 years (95% CI)	Hazard ratio (95% CI)	P*
OS	NFκB intensity	Negative (RL) vs positive	22/26	50.0 (29.9, 67.2)	0.88 (0.50, 1.55)	0.66
			27/29	65.5 (45.4, 79.7)
	NFκB intensity	Negative (RL) vs positive	22/26	50.0 (29.9, 67.2)	0.75 (0.42, 1.35)	0.34
			27/29	65.5 (45.4, 79.7)
	Treatment arm	RT alone (RL) vs RT + HT	26/30	56.7 (37.3, 72.1)	0.75 (0.40, 1.40)	0.37
			23/25	60.0 (38.4, 76.1)
	Age, years	< 75 (RL) vs ≥ 75	36/42	54.8 (38.7, 68.3)	1.72 (0.86, 3.43)	0.13
			13/13	69.2 (37.3, 87.2)
	Combined Gleason score	2–6 (RL) vs 7–10	11/13	84.6 (51.2, 95.9)	2.51 (1.19, 5.31)	0.02
			38/42	50.0 (34.2, 63.9)
	Clinical stage	T2 (RL) vs T3	12/14	64.3 (34.3, 83.3)	1.33 (0.66, 2.67)	0.43
			37/41	56.1 (39.7, 69.6)
DFS	NFκB intensity	Negative (RL) vs positive	25/26	26.9 (11.9, 44.5)	1.09 (0.62, 1.90)	0.77
			29/29	13.8 (4.3, 28.6)
	NFκB intensity	Negative (RL) vs positive	25/26	26.9 (11.9, 44.5)	1.00 (0.56, 1.79)	0.99
			29/29	13.8 (4.3, 28.6)
	Treatment arm	RT alone (RL) vs RT + HT	30/30	13.3 (4.2, 27.8)	0.41 (0.22, 0.74)	0.003
			24/25	28.0 (12.4, 46.0)
	Age, years	< 75 (RL) vs ≥ 75	41/42	19.0 (8.9, 32.0)	0.75 (0.39, 1.47)	0.41
			13/13	23.1 (5.6, 47.5)
	Combined Gleason score	2–6 (RL) vs 7–10	13/13	46.2 (19.2, 69.6)	2.48 (1.15, 5.35)	0.02
			41/42	11.9 (4.4, 23.6)
	Clinical stage	T2 (RL) vs T3	14/14	21.4 (5.2, 44.8)	0.75 (0.37, 1.54)	0.44
			40/41	19.5 (9.2, 32.7)
OS	CXCR4 intensity	Negative (RL) vs positive	24/28	71.4 (50.9, 84.6)	1.46 (0.85, 2.51)	0.17
			31/35	57.1 (39.3, 71.5)
	CXCR4 intensity	Negative (RL) vs positive	24/28	71.4 (50.9, 84.6)	1.45 (0.83, 2.55)	0.19
			31/35	57.1 (39.3, 71.5)
	Treatment arm	RT alone (RL) vs RT + HT	29/33	60.6 (42.0, 74.9)	0.66 (0.37, 1.19)	0.17
			26/30	66.7 (46.9, 80.5)
	Age, years	< 75 (RL) vs ≥ 75	40/48	60.4 (45.2, 72.6)	1.50 (0.81, 2.75)	0.20
			15/15	73.3 (43.6, 89.1)
	Combined Gleason score	2–6 (RL) vs 7–10	13/16	87.5 (58.6, 96.7)	1.77 (0.90, 3.48)	0.10
			42/47	55.3 (40.1, 68.1)
	Clinical Stage	T2 (RL) vs T3	11/13	69.2 (37.3, 87.2)	1.69 (0.81, 3.54)	0.17
			44/50	62.0 (47.1, 73.8)
DFS	CXCR4 intensity	Negative (RL) vs positive	28/28	21.4 (8.7, 37.8)	1.08 (0.65, 1.79)	0.77
			34/35	14.3 (5.2, 27.7)
	CXCR4 intensity	Negative (RL) vs positive	28/28	21.4 (8.7, 37.8)	1.06 (0.62, 1.79)	0.82
			34/35	14.3 (5.2, 27.7)
	Treatment arm	RT alone (RL) vs RT + HT	33/33	12.1 (3.8, 25.5)	0.40 (0.23, 0.70)	0.001
			29/30	23.3 (10.3, 39.4)
	Age, years	< 75 (RL) vs ≥ 75	47/48	16.7 (7.8, 28.4)	0.77 (0.42, 1.42)	0.41
			15/15	20.0 (4.9, 42.4)
	Combined Gleason score	2–6 (RL) vs 7–10	16/16	37.5 (15.4, 59.8)	2.43 (1.23, 4.79)	0.01
			46/47	10.6 (3.9, 21.3)
	Clinical Stage	T2 (RL) vs T3	13/13	23.1 (5.6, 47.5)	0.93 (0.46, 1.86)	0.83
			49/50	16.0 (7.5, 27.4)

^*P-values from Chi-square test using Cox Proportional Hazards Model. A hazard ratio is defined as the ratio of the estimated hazard for those with a variable value 1 to the estimated hazard for those with a variable value 0. HT, hormone therapy.

Fig. 2.

OS (A) and DFS (B) by dichotomized NFκB intensity score (Kaplan–Meier Graph). HR, hazard ratio.

Fig. 3.

OS (A) and DFS (B) by dichotomized CXCR4 intensity score (Kaplan–Meier Graph). HR, hazard ratio.

Table 4.

Cox proportional hazards regression model by dichotomized CXCR4 intensity score and dichotomized NFκB

Outcome	Covariate	Comparison	Fail/total	Survival rate at 5 years (95% CI)	Hazard ratio (95% CI)	P*
OS	Both CXCR4 and NFκB	At least one negative (RL) vs both positive	28/33	54.5 (36.3, 69.6)	1.14 (0.62, 2.10)	0.67
			17/18	66.7 (40.4, 83.4)
	Both CXCR4 and NFκB	At least one negative (RL) vs both positive	28/33	54.5 (36.3, 69.6)	0.91 (0.48, 1.73)	0.77
			17/18	66.7 (40.4, 83.4)
	Treatment arm	RT alone (RL) vs RT + HT	23/27	55.6 (35.2, 71.8)	0.73 (0.37, 1.44)	0.37
			22/24	62.5 (40.3, 78.4)
	Age, years	< 75 (RL) vs ≥ 75	32/38	55.3 (38.3, 69.3)	1.86 (0.90, 3.82)	0.09
			13/13	69.2 (37.3, 87.2)
	Combined Gleason score	2–6 (RL) vs 7–10	10/12	83.3 (48.2, 95.6)	2.08 (0.95, 4.53)	0.07
			35/39	51.3 (34.8, 65.5)
	Clinical stage	T2 (RL) vs T3	10/12	66.7 (33.7, 86.0)	1.61 (0.75, 3.44)	0.22
			35/39	56.4 (39.6, 70.2)
DFS	Both CXCR4 and NFκB	At least one negative (RL) vs both positive	32/33	24.2 (11.4, 39.6)	1.11 (0.62, 2.00)	0.73
			18/18	16.7 (4.1, 36.5)
	Both CXCR4 and NFκB	At least one negative (RL) vs both positive	32/33	24.2 (11.4, 39.6)	1.06 (0.58, 1.94)	0.85
			18/18	16.7 (4.1, 36.5)
	Treatment arm	RT alone (RL) vs RT + HT	27/27	14.8 (4.7, 30.4)	0.39 (0.21, 0.73)	0.003
			23/24	29.2 (13.0, 47.6)
	Age, years	< 75 (RL) vs ≥ 75	37/38	21.1 (9.9, 35.0)	0.82 (0.41, 1.62)	0.56
			13/13	23.1 (5.6, 47.5)
	Combined Gleason score	2–6 (RL) vs 7–10	12/12	50.0 (20.8, 73.6)	2.60 (1.17, 5.76)	0.02
			38/39	12.8 (4.7, 25.2)
	Clinical stage	T2 (RL) vs T3	12/12	25.0 (6.0, 50.5)	0.85 (0.40, 1.82)	0.67
			38/39	20.5 (9.6, 34.2)

^*P-values from Chi-square test using Cox Proportional Hazards Model. A hazard ratio is defined as the ratio of the estimated hazard for those with a variable value 1 to the estimated hazard for those with a variable value 0. A hazard Ratio of 1 indicates no difference between two subgroups. HT, hormone therapy.

Discussion

There are an increasing number of potential biomarkers and therapeutic targets for prostate cancer. To enter this territory, understanding the biological and molecular processes of prostate carcinogenesis and the natural history of the disease is essential. The results of the present study indicate NFκB and CXCR4 are expressed at moderate-to-high levels in about half of locally advanced organ-confined prostate cancers (per eligibility for RTOG 8610 trial). Moreover, the data suggests that CXCR4 is co-expressed with NFκB in about one third of patients, and supports the notion that CXCR4 is regulated by NFκB in prostate cancer. It is important to note that the present retrospective study was performed on prostate samples from the RTOG 8610 trial and patients were hormone naïve at the time of surgery. It is possible that individual and co-expression of NFκB/CXCR4 changes over time and possibly because of androgen withdrawal and development of castrate resistance. The answer to this question is outside the scope of the present study.

Although both NFκB and CXCR4 were expressed in a significant number of patients with organ-confined prostate cancer, neither predicted clinical outcomes in the present analysis. This might be due to the small sample size, half the patients receiving radiation alone and the other half radiation plus androgen-deprivation therapy, and reliance of assessing NFκB cytoplasmic staining rather than nuclear staining (i.e. NFκB’s site of action) Nevertheless, the present study further supports the increasing data that NFκB and CXCR4 are abnormally expressed on prostate cancer cells. Existing studies already suggest that individual NFκB and CXCR4 expression may portend worse prognosis in prostate cancer and increased NFκB activity may also predispose to hormonal and chemotherapy resistance [20,25]. Their utility as predictive and/or prognostic factors should be explored in future clinical trials. Also, because drugs are now available to target these proteins and promising preclinical data exists in prostate cancer [5,21], studies in larger prospective studies to definitively determine the utility of NFκB and CXCR4 as predictive biomarkers and therapeutic targets are warranted.

Understanding the interplay between NFκB, CXCR4 and other factors will be critical to moving the science onto the clinical arena where therapeutic agents can be developed. CXCR4 expression is associated with the activity of interleukin 8, matrix metalloproteinase 9 and vascular endothelial growth factor [26,27], which each play roles in carcinogenesis such as angiogenesis and tumour invasion. However, together with CXCR4, these factors are also often regulated by NFκB, which activates their genes upon stimulus. In turn, NFκB overactivity may occur as a result of changes in the phosphatidylinositol-3-kinase (PI3K)/protein kinase B (PKB/Akt) pathway [28,29] and this may again widen the possibilities for targets for drug development. In conjunction with the data presented here, it would then seem rationale to target the upstream regulator such as NFκB rather than CXCR4 alone.

In conclusion, NFκB and CXCR4 are co-expressed in a about one third of patients with prostate cancer and higher levels of NFκB may be associated with higher levels of CXCR4. The exact nature of their interaction and their combined effects upon prostate cancer growth and progression is still unclear and requires further research, as does their utility as predictive and prognostic factors.

Abbreviations

DAB

diaminobenzidine

DFS

disease-free survival

HRP

horseradish peroxidase

IκB

inhibitoryκB proteins

NFκB

nuclear factor-κB

OS

overall survival

RL

reference level

RT

radiation therapy

RTOG

Radiation Therapy Oncology Group

STAD

short-term neoadjuvant and concurrent androgen deprivation

TBS

Tris buffer saline