The expression pattern of Ku correlates with tumor radiosensitivity and disease free survival in patients with rectal carcinoma

Abstract

BACKGROUND

Preoperative radiotherapy reduces the rate of local recurrence and improves the chance of survival in patients with resectable, advanced rectal carcinoma. However, because not all tumors respond similarly to radiation, sorting out suitable patients is required to irradiate tumors rationally. The authors examined the possible role of Ku protein in determining tumor radiosensitivity and disease free survival in patients with rectal carcinoma.

METHODS

The authors studied 96 patients with advanced rectal carcinoma. In preradiation biopsy specimens of tumor samples, the number of cells that were stained positive for Ku protein was evaluated by immunohistochemistry. The expression pattern of Ku protein was examined for an association between tumor radiosensitivity (which was determined according to T classification downstaging, complete pathologic response, or Response Evaluation Criteria in Solid Tumors) and disease free survival.

RESULTS

There was a high degree of correlation between the percentage of cells that expressed the 70-kDa Ku protein (Ku70) and the 86-kDa Ku protein (Ku86) in the tumor sections (correlation coefficient = 0.85; P < 0.001). The expression pattern of Ku protein was correlated not only with tumor radiosensitivity but also with disease free survival. Pathologic TMN classification, histopathologic grade, and Ku70 expression were significant prognostic variables for disease free survival in a multivariate analysis (P = 0.0031, P = 0.030, and P = 0.023, respectively).

CONCLUSIONS

Ku70 and Ku86 raise the predictive possibility of tumor radiosensitivity. Ku may be a useful parameter for selecting patients with rectal carcinoma for preoperative radiotherapy. Cancer 2002;95:1199–205. © 2002 American Cancer Society.

DOI 10.1002/cncr.10807

Preoperative radiation as adjuvant therapy for patients with locally advanced rectal carcinoma has gained wide use, because radiation can decrease the tumor bulk and may change it from a fixed and unresectable tumor into a mobile and resectable tumor.1–3 Furthermore, in the Swedish Rectal Cancer Trial, preoperative radiotherapy reduced the rate of local recurrence and improved the chance of survival in patients with resectable rectal carcinoma.4, 5 However, because not all tumors respond similarly to radiation, sorting out suitable patients is required to irradiate tumors rationally. Recent studies have shown that the presence of mutated p53, which is involved in the regulation of the cell cycle and apoptosis, usually predicts tumor resistance to radiation.6–12

The immunohistochemical analysis of DNA repair enzymes, one of which is the Ku protein13—a heterodimer comprised of a 70-kilodalton (kDa) protein (Ku70) and a 86-kDa protein (Ku86) and is involved in the recognition or repair of radiation-induced DNA damage—may have potential as a predictive assay for tumor radiosensitivity.14, 15 In patients with carcinoma of the cervix, the expression pattern of Ku protein reportedly was correlated with tumor radiosensitivity and survival.16 In this study, we examined the possible role of Ku protein in determining tumor radiosensitivity and disease free survival (DFS) in patients with rectal carcinoma.

MATERIALS AND METHODS

Patients

A total of 180 consecutive patients with rectal carcinoma underwent preoperative radiation and surgical resection at the Department of Surgical Oncology at the University of Tokyo between 1985 and 2000. Among the 180 patients, 96 patients who were diagnosed with adenocarcinoma were studied, for whom both preradiation biopsy specimens and postirradiation resected specimens were available. All tumors were locally advanced (T3 or T4), as determined by endoscopic ultrasonography or computed tomography (CT) scan. No distant metastases of the lung were determined by plain chest X-ray or CT scan, and both the liver and para-aortic lymph nodes were negative for metastases on ultrasonography and CT scans. The total dose of radiation delivered through a linear accelerator was 50 grays (Gy) in all patients, usually in doses of 2 Gy at each treatment, 5 times per week. All patients were followed, and follow-up ranted from 1 month to 161 months (mean, 41.1 months; median, 27 months). Informed consent was obtained from all patients who participated in this study.

Immunohistochemical Staining of Ku70 and Ku86

The streptoavidin-biotin peroxidase complex technique was used for staining sections. Formalin fixed, paraffin embedded sections (3 μm thick) of preradiation biopsy specimens were dewaxed for 15 minutes in xylene and hydrated by passage through a graded ethanol series to tap water. Antigen retrieval was performed routinely by incubation in target retrieval solution (Dako, Tokyo, Japan) at 97 °C for 20 minutes. After blocking with 0.3% hydrogen peroxide in methanol, the sections were blocked for another 30 minutes with 10% normal goat serum. Anti-Ku70 and anti-Ku86 rabbit polyclonal antisera, which have been described elsewhere,17 or Tris-buffered saline (TBS) for control sections was added. They were used as 1:75 dilutions in TBS. After incubation at 4 °C for 18 hours, sections were incubated with a second biotinylated antibody of goat antirabbit specificity (Nichirei, Tokyo, Japan) for 20 minutes. Then, avidin-peroxidase reagent (Nichirei) was applied for 10 minutes. All sections were counterstained with hematoxylin. Reactivity was visualized with 3,3′-diaminobenzidine as the substrate, yielding a brown reaction product. The sections were then dehydrated through a graded ethanol series to xylene and mounted.

Quantification of Ku70 and Ku86 Expression Pattern in Tumor Sections Using Light Microscopy

In preradiation biopsy specimens of tumor samples, the numbers of cells that were stained positive for Ku70 and Ku86 were determined by scoring 10 microscopic fields of 100 tumor cells each, without prior knowledge of radiosensitivity or treatment outcome, and determining the percentages of cells that were positive for Ku70 expression and Ku86 expression. We defined the expression pattern of the Ku protein as high or low as follows; 1) Ku70 High and Ku86 High: the percentage of cells positive for Ku70 or Ku86 was equal to or greater than the median; and 2) Ku70 Low and Ku86 Low: the percentage of cells positive for Ku70 or Ku86 was less than the median.

Tumor Radiosensitivity

Tumor radiosensitivity was determined by three methods. First, T classification downstaging was measured by differences in T classification before and after preoperative radiotherapy. Second, pathologic responses were divided into two categories: complete pathologic responses or microscopic residual disease. Finally, tumors were categorized according to the Response Evaluation Criteria in Solid Tumors.18 Both complete responses and partial responses were defined as objective responses. Stable disease and progressive disease were defined as nonresponses.

Statistical Analysis

We tested correlations between quantitative variables by establishing nonparametric linear regression. Associations between the expression pattern of Ku protein and tumor radiosensitivity were determined with a Fisher exact test. Actuarial survival curves were calculated according to the Kaplan–Meier method.19 The probabilities of DFS were determined by univariate log-rank analysis.20 Multivariate analyses used the Cox proportional hazards model21 of prognostic factors for DFS and a stepwise variable selection process.22 A significance level of 0.05 was used throughout.

RESULTS

Patient Characterization

There were 68 men and 28 women, ranging in age from 28 years to 84 years (mean, 59.6 years; median, 60 years). Seventy-five tumors were well differentiated, 16 tumors were moderately differentiated, and 5 tumors were mucinous adenocarcinomas. According to pathologic TMN classification, it was determined after surgery that 7 patients had Stage 0 disease, 14 patients had Stage I disease, 38 patients had Stage II disease, and 37 patients had Stage III disease.

Immunostaining

TBS was used as a negative control. No staining was seen in any of the sections when TBS was used. Staining for Ku70 and Ku86 was predominantly nuclear (Fig. 1). The percentage of cells that expressed Ku70 and Ku86 in different tumors ranged from 2.3% to 97.5% for Ku70 and from 2.1% to 94.8% for Ku86 (Table 1). The median values for Ku70 and Ku86 were 68.4 and 69.2, respectively. Similar staining patterns for both Ku70 and Ku86 could be seen in the adjacent sections (Fig. 1). There was a high degree of correlation between the percentage of cells expressing Ku70 and Ku86 in the tumor sections (r = 0.85; P < 0.001) (Fig. 2).

Figure 1. Ku immunostaining. (A) Hematoxylin and eosin stain. (B) Immunostaining for the 70-kilodalton (kDa) Ku protein (Ku70). (C) Immunostaining for the 86-kDa Ku protein (Ku86). Staining for Ku70 and Ku86 was predominantly nuclear, and a similar staining pattern for both Ku70 and Ku86 could be seen in adjacent sections (B,C: streptoavidin-biotin peroxidase complex method; original magnification ×400).

Table 1. Biologic Parametersa

Expression	No.	Mean ± SD (%)	Median (%)	Range (%)
SD. standard deviation. a The levels of Ku70 and Ku86 expression were determined by scoring 10 microscope fields of 100 tumor cells each and determining the percentage of cells positive for Ku70 and Ku86 expression. Values were expressed as mean percentages of the 10 fields scored. The mean, median, and range of the 96 tumors scored are given.
Ku70	96	61.7 ± 30.2	68.4	2.3–97.5
Ku86	96	63.0 ± 27.5	69.2	2.1–94.8

Figure 2. Expression of the 70-kilodalton (kDa) Ku protein (Ku70) and the 86-kDa Ku protein (Ku86) in adjacent sections of 96 rectal carcinomas. Open circles represent the mean number of cells that stained positive for Ku70 or Ku86 in 10 microscopic fields of 100 tumor cells each.

Relations between Ku Expression and Tumor Radiosensitivity

Relations between the expression patterns of Ku protein in preradiation biopsy specimens and tumor radiosensitivity are summarized in Tables 2–4. In an analysis of Ku70, there was a correlation between the expression pattern of Ku70 protein and tumor radiosensitivity in T classification downstaging, pathologic response, and the Response Evaluation Criteria in Solid Tumors. Tumors that had a high percentage of Ku70 positive cells tended to be radioresistant. Conversely, for Ku86, only in the Response Evaluation Criteria in Solid Tumors was the expression pattern of Ku86 protein correlated with tumor radiosensitivity.

Table 2. Relations between the Expression Patterns of Ku Protein and Tumor Radiosensitivity: T Classification Downstaging of Patients with Rectal Carcinoma in Relation to Ku70/Ku86 Staininga

Staining	Downstaged	Nonresponse	P value
a Downstaged: T classification after irradiation was decreased compared with T classification before irradiation: nonresponse: T classification after irradiation was not decreased compared with T classification before irradiation.
Ku70
Low	24	24	—
High	13	35	0.035
Ku86
Low	21	27	—
High	16	32	0.40

Table 3. Relations between the Expression Patterns of Ku Patterns and Tumor Radiosensitivity. Pathologic Response of Patients with Rectal Carcinoma in Relation to Ku70/Ku86 Staininga

Staining	Complete pathologic response	Microscopic residual disease	P value
a Complete pathologic response: no residual tumor cells were seen in resected tissue; microscopic residual disease: residual tumor cells were seen in resected tissue.
Ku70
Low	7	41	—
High	0	48	0.012
Ku86
Low	6	42	—
High	1	47	0.11

Table 4. Relations between the Expression Patterns of Ku Protein and Tumor Radiosensitivity. Objective Response of Patients with Rectal Carcinoma in Relation to Ku70/Ku86 Staininga

Staining	Objective response	Nonresponse	P value
a Objective response: both complete responses and partial responders were included in this category; nonresponse: both patients with stable disease and patients with progressive disease were included in this category.
Ku70
Low	37	11
High	20	28	< 0.001
Ku86
Low	35	13	—
High	22	26	0.012

Relations between Ku Expression Pattern and Clinicopathologic Features

Table 5 summarizes the distribution of patients regarding clinicopathologic features. No significant correlations were seen between Ku expression patterns and the clinical features, except between Ku70 expression and pathologic T classification.

Table 5. Ku70 and Ku86 Expression in Relation to Clinical Parameters

Parameter	Ku70			Ku86			Total
	Low	High	P value	Low	High	P value
Wel/Mod; well or moderately differentiated adenocarcinoma; Muc: mucinous adenocarcinoma. a Values shown for age are the mean years ± standard deviation.
Age (ys)a	58.5 ± 10.4	60.7 ± 11.7	0.33	57.5 ± 10.3	61.7 ± 11.5	0.063	59.6 ± 11.0
Male	35	33	—	34	34	—	68
Female	13	15	0.82	14	14	0.99	28
Clinical T classification
T3	42	41	—	44	39	—	83
T4	6	7	0.99	4	9	0.23	13
pT
pT0,1,2	19	8	—	18	9	—	27
pT3,4	29	40	0.022	30	39	0.070	69
pN
pN0	31	28		33	26	—	59
pN1,2	17	20	0.68	15	22	0.21	37
pTMN stage
0–II	31	28	—	33	26	—	59
III	17	20	0.68	15	22	0.21	37
Histology
Wel/Mod	46	45	—	47	44	—	91
Muc	2	3	0.99	1	4	0.36	5
Lymphatic invasion
Negative	38	41	—	38	41	—	79
Positive	10	7	0.59	10	7	0.59	17
Venous invasion
Negative	24	21	—	22	23	—	45
Positive	24	27	0.68	26	25	0.99	51

Survival

Log-rank analysis was performed to determine whether Ku70 or Ku86 expression patterns could be used as prognostic indicators for the response of patients with rectal carcinoma to radiotherapy (Figs. 3, 4). DFS levels for patients with tumors with Ku70 High versus Ku70 Low expression were 42.1% and 77.9%, respectively. The difference reached statistical significance (P = 0.0057). DFS levels in patients with Ku86 High versus Ku86 Low expression were 48.8% and 75.2%, respectively, and these differences also reached statistical significance (P = 0.022). The DFS level for patients with pathologic TMN Stage 0, I, and II disease was 77.6%, and the DFS level for patients with pathologic TMN Stage III disease was 46.1%. The DFS level for patients with well or moderately differentiated adenocarcinomas was 67.7%; and the DFS level for patients with mucinous adenocarcinomas was 20.0%. Both pathologic TMN classification and histopathologic grade reached statistical significance (P = 0.0016 and P = 0.0005, respectively).

Figure 3. Disease free survival in patients with rectal carcinoma after radiotherapy was compared with expression levels of the 70-kilodalton Ku protein (Ku70). Survival is shown after surgery among patients who were grouped according to low Ku70 tumor expression (Ku70 Low) or high Ku70 tumor expression (Ku70 High).

Figure 4. Disease free survival in patients with rectal carcinoma after radiotherapy was compared with expression levels of the 86-kilodalton Ku protein (Ku86). Survival is shown after surgery among patients who were grouped according to low Ku86 tumor expression (Ku86 Low) or high Ku86 tumor expression (Ku86 High).

We performed multivariate analyses to identify the independent factors that affected DFS. The results are shown in Table 6. Pathologic TMN classification, histopathologic grade, and Ku70 expression were significant prognostic variables for DFS in the multivariate analysis (P = 0.0031, P = 0.030, and P = 0.023, respectively).

Table 6. Multivariate Analysesa

Factor	Univariate	Multivariate
		Hazard rate	95%CI	P value
Hist: histopathologic staging; 95% CI: 95% confidence interval. a Results are shown of univariate analyses using the log-rank tests and multivariate analyses using the Cox proportional hazards model of prognostic factors for disease free survival. A stepwise variable selection process was used, and the final model is shown.
pTMN (0,I,II vs. III)	0.0016	8.7	0.14–0.67	0.0031
Hist (well, moderate vs. mucinous)	0.0005	4.7	0.09–0.89	0.030
Ku70 (low vs. high)	0.0057	5.1	0.16–0.88	0.023

DISCUSSION

Mitotic or clonogenic cell death is considered the major mechanism by which most solid tumors respond to clinical radiotherapy.23 The correlation between the amount of mitotic cell death, DNA lesion induction, and chromosomal aberrations suggests that this pattern of cell death probably results from failure of cells to completely or accurately repair DNA damage.23 Among different types of DNA damage, DNA double-strand breaks are the major lethal lesions induced by ionizing radiation.24 A nonhomologous, end-joining pathway is an important repair pathway for DNA damage.25–27 There are a number of observations suggesting that DNA protein kinase (DNA-PK) has a specific role in the nonhomologous, end-joining pathway.16 It is comprised of a catalytic subunit, DNA-PKcs, and the regulatory subunit Ku.13 The latter is a heterodimer comprised of 70-kDa (Ku70) and 86-kDa (Ku86) proteins.

Wilson et al. reported that there was a high degree of correlation between the percentage of cells expressing Ku70 and Ku86 in tumor sections.16 In the current study, a similar pattern of staining for Ku70 and Ku86 could be seen in adjacent sections. This supports the interdependence of the two heterodimer components.28

There was a correlation between the expression pattern of Ku protein in preradiation biopsy specimens of tumor samples and tumor radiosensitivity. Tumors with a high percentage of Ku70 and Ku86 positive cells tended to be radioresistant. Conversely, tumors with a low percentage of Ku70 and Ku86 positive cells tended to be radiosensitive. Therefore, Ku70 and Ku86 raise the predictive possibility for tumor radiosensitivity. However, the accuracy of the expression pattern of Ku70 and Ku86 was comparatively low, ranging from 55% to 68%. There are three possible reasons: 1) There are limitations in assay sensitivity by immunohistochemistry (a subtle difference in expression patterns of Ku70 and Ku86 that may influence the radiosensitivity of cells may not be detected); 2) the expression patterns of Ku70 and Ku86 do not necessarily reflect their function; and 3) the diversity of regulation of radiation sensitivity. The response of tumor cells to ionizing radiation includes activation of DNA repair pathways, cell cycle checkpoints, partial pressure oxygen of the tumor, and so on.23 Of these, only DNA repair pathways do not determine the effect of irradiation. This finding suggests that DNA-PK is involved in determining the outcome of patients with malignant disease through its involvement in repair of DNA double-strand breaks but that other factors also are important.

A link between Ku expression pattern and ionizing radiation sensitivity is supported further by the significantly higher survival of patients with Ku70 Low and Ku86 Low expression in univariate analysis (P = 0.0057 and P = 0.022, respectively). The poorer prognosis of patients with Ku70 High and Ku86 High expression suggests that tumor cells with the ability to repair damaged DNA are more likely to survive and proliferate after irradiation. Furthermore, pathologic TMN classification, histopathologic grade, and Ku70 expression were predictive factors in multivariate analysis (P = 0.012, P = 0.041, and P = 0.014, respectively). Pathologic TMN classification and histopathologic grade have remained the most important predictive variables for survival;29 however, these variables are determined only after surgery. We can evaluate the expression pattern of Ku protein before patients undergo preoperative radiation. Ku, therefore, may be a useful parameter for selecting patients with rectal carcinoma for preoperative radiotherapy.

Acknowledgements

The authors thank Shinsuke Saito, M.D., and Soichiro Ishihara, M.D., for their excellent technical advice.