Transketolase-like protein 1 expression predicts poor prognosis in colorectal cancer

ABSTRACT

Background: Transketolase-like protein 1 (TKTL1) is an isoform of tranketolase, a key protein in a cancer cell's glucose metabolism that causes rapid cell growth and controls the non-oxidative part of the pentose phosphate pathway (PPP). Its overexpression occurs in several human cancer types. Our purpose was to study whether TKTL1 expression in colorectal cancer tissue associates with these patients’ prognosis. Methods: We collected retrospectively patient data and tissue samples from 840 colorectal cancer patients treated at Helsinki University Hospital, then stained tumor tissue microarrays for TKTL1 by immunohistochemistry, and compared immunohistochemical tissue expression with clinico-pathological parameters and survival. Results: High expression of TKTL1 associated with high Dukes stage, non-mucinous adenocarcinoma, and left-sided disease. Patients with high TKTL1 expression had poorer prognosis than those with low expression, with a 5-year disease-specific survival of 55.7% vs. 62.7%. Conclusion: We show that high TKTL1 in tumor tissue can lead to poor survival in colorectal cancer. TKTL1 thus can serve as a candidate marker for identifying patients at risk of recurrent disease.

KEYWORDS:

• Colon cancer
• colorectal cancer
• immunohistochemistry
• prognosis
• rectal cancer
• TKTL1
• tumor marker

Abbreviation

TKTL1

=

transketolase-like protein 1

Introduction

Cancer cells show an increased glycolysis even in aerobic environments via the pentose phosphate pathway (PPP), a phenomenon known as the Warburg effect. ¹ The non-oxidative part of the PPP is controlled by transketolase enzyme reactions.² Transketolase-like protein 1 (TKTL1) is an isoform of transketolase that is shown to cause rapid tumor-cell growth. Suppression of TKTL1, on the other hand, reduces glucose consumption, with reduced lactic acid production.³

In human colon carcinoma, TKTL1-suppressed cells grow more slowly than do controls.³ Overexpression of TKTL1 has been demonstrated in non-small-cell lung carcinoma, breast carcinoma, head and neck squamous cell carcinoma, follicular as well as papillary thyroid carcinoma, and in carcinomas of the prostate, pancreas, ovary, endometrium, cervix, rectum, and kidney.^4-10 In bladder cancer, its overexpression occurs only in invasive tumors, whereas non-muscle-invading tumors show no TKTL1 positivity. Moreover, noninvasive colon carcinomas (pTis tumors) are TKTL1 negative.⁴ TKTL1 overexpression predicts, in locally advanced rectal cancer treated with preoperative chemoradiation, poor patient survival and tumor progression.¹¹ In many other cancer forms too, such as colon, ocular, non-small cell lung cancer, and oral squamous cell cancer, TKTL1 overexpression predicts poor outcome.^4,7,12-14

Colorectal cancer is the third most common cancer in men and the second most common in women globally. Some 1.4 million new cases are diagnosed annually, causing annually 694 000 deaths. Of the cases, 55% occur in the developed world.¹⁵ Those with UICC stage I to II disease affecting only the mucosa and muscular layer of the colon or rectum with no lymph node metastasis are, after surgical removal of the bowel, usually offered follow-up only. However, among stage II patients, some 20% die of their cancer.¹⁶ The question is: who are those patients with stage II disease with high risk of developing recurrent disease and to whom should we offer adjuvant therapy.

The aim of our study was to explore the prognostic impact of TKTL1 tumor expression on survival in colorectal cancer.

Results

Immunohistochemistry

Of 840 patients, a total of 733 could be evaluated. Of these, 105 (12.5%) were scored as negative (score 0), 268 (31.9%) weakly positive (score 1), 235 (28.0%) moderately positive (score 2), and 125 (14.9%) strongly positive (score 3; Fig. 1). For further analyses, those 373 with negative and weak TKTL1 expression were grouped as the low expression group, and those 360 with moderate and strong expression as high expression.

Figure 1. TKTL1 immunohistochemical staining of colorectal cancer samples. Samples were scored for cytoplasmic intensity by a 4-grade scale as follows, (A) negative (B) mild (C) intermediate (D) strong. Original magnification was 400X.

Association analysis

High expression of TKTL1 associated with Dukes stage B to D (P = 0.002), with adenocarcinoma (P ≤ 0.001), and with left-sided disease (P = 0.044, chi-square test). We found no association between TKTL1 and age, gender, or WHO grade, nor did any difference emerge in TKTL1 expression in the colon versus rectum (Table 1).

Table 1. Association of TKTL1 with clinicopathologic variables in 733 colorectal cancer patients.

		TKTL-1 low (0-1) expression		TKTL-1 high (2-3) expression
Clinicopathological variable	n	n	%	n	%	P-value*
Age, years
< 65	312	164	52.6	148	47.4	0.434
≥ 65	421	209	49.6	212	50.4
Gender
Male	412	210	51.0	202	49.0	0.959
Female	321	163	50.8	158	49.2
Grade (WHO)
1	25	16	64.0	9	36.0	0.517
2	511	261	51.1	250	48.9
3	172	84	48.8	88	51.2
4	22	10	45.5	12	54.5
Dukes classification
A	108	73	67.6	35	32.4	0.002
B	257	126	49.0	131	51.0
C	203	95	46.8	108	53.2
D	165	79	47.9	86	52.1
Histologic type
Adeno	659	321	48.7	338	51.3	<0.001
Adenomucinous	73	52	71.2	21	28.8
Location
Colon	384	197	51.3	187	48.7	0.813
Rectum	349	176	50.4	173	49.6
Side
Right	202	115	56.9	87	43.1	0.044
Left	531	258	48.6	273	51.4

Abbreviations TKTL1 = transketolase like protein 1

Survival analysis

Patients with high TKTL1 expression had a poor prognosis. Disease-specific survival (DSS) at 5 y for all patients in this study was 58.9% (95% CI 55.4-62.4). For those with low TKTL1 expression 5-year DSS was 62.7% (95% CI 57.6–67.8), compared to those with high expression at 55.7% (95% CI 50.4–61.0; P = 0 .024, log-rank test; Fig. 2). In subgroup analysis, TKTL1 served as a prognostic factor for patients under 65 (P = 0.015), patients with adenocarcinoma (P = 0.011), and patients with a high proliferation index (>10 % Ki-67 positivity) (P = 0.029, log-rank test; Table 2).

Table 2. Kaplan-Meier analysis for disease-specific survival stratified for subgroups of colorectal cancer patients (P-value for logrank test).

	5-year cumulative survival (95% CI)
Subgroup	All patients	TKTL-1 low	TKTL-1 high	P-value
TKTL-1	58.9 (55.4–62.4)	62.7 (57.6–67.8)	55.7 (50.4–61.0)	0.024
Age, years
< 65	63.8 (58.7–68.9)	68.5 (61.2–75.6)	56.1 (48.1–64.1)	0.015
≥ 65	55.0 (50.3–59.7)	58.0 (50.7–65.3)	55.3 (48.2–62.4)	0.477
Gender
Male	60.1 (55.4–64.8)	63.1 (56.2–70.0)	57.8 (50.5–65.1)	0.113
Female	57.3 (52.2–62.4)	62.0 (54.1–69.8)	53.1 (45.1–61.1)	0.123
Grade
1–2	62.5 (58.4–66.6)	65.5 (59.6–71.4)	60.4 (54.1–66.7)	0.102
3–4	18.2 (11.3–25.1)	53.2 (42.6–63.8)	42.4 (32.2–52.6)	0.147
Dukes classification
A	91.3 (86.2–96.4)	91.3 (84.6–98.0)	93.3 (84.3–1.02)	0.570
B	77.4 (72.3–82.5)	79.0 (71.6–86.4)	78.3 (70.9–85.7)	0.797
C	55.1 (48.2–62.0)	59.8 (49.2–70.4)	54.1 (44.3–63.9)	0.259
D	9.8 (5.3–14.3)	5.9 (0.2–11.6)	8.2 (1.9–14.5)	0.542
Histologic type
Adeno	59.6 (55.9–63.3)	63.9 (58.4–69.4)	56.3 (50.8–61.8)	0.011
Adenomucinous	52.6 (41.8–63.4)	55.5 (41.6–69.4)	45.0 (23.2–66.8)	0.929
Side
Right	58.3 (51.6–65.0)	62.9 (53.7–72.1)	52.0 (40.8–63.2)	0.063
Left	59.0 (54.9–63.1)	62.4 (56.1–68.7)	56.8 (50.7–62.9)	0.127
Location
Colon	58.8 (53.9–63.7)	62.2 (55.1–69.3)	54.9 (47.5–62.3)	0.098
Rectum	58.8 (53.7–63.9)	63.2 (55.5–70.8)	56.6 (48.8–64.4)	0.137
Ki-67
≤ 10%	55.0 (49.1–60.9)	54.2 (45.8–62.6)	54.3 (45.3–63.3)	0.701
> 10%	62.1 (54.4–69.7)	69.2 (58.6–79.8)	55.0 (43.6–66.4)	0.029

Abbreviations: TKTL1 = ttransketolase -like protein 1; CI = confidence interval

Figure 2. Disease-specific survival according to TKTL1 expression in colorectal cancer patients. Prognosis was better in those with low expression of TKTL1, and worse in those with high expression of TKTL1 (p = 0 .024)

In multivariable survival analysis, patients’ age (hazard ratio (HR) 1.92, 95% CI 1.53-2.42, P < 0.001), Dukes classification, and WHO grade were independent risk factors, but TKTL1 showed no independent prognostic value (Cox regression analyses; Table 3).

Table 3. Cox regression analysis for disease-specific survival of colorectal cancer patients.

	Univariable survival analysis			Multivariable survival analysis
Variable	Hazard ratio	95% CI	P-value	Hazard ratio	95% CI	P-value
Age, years
< 65	1.00			1.00
≥ 65	1.41	1.14–1.73	0.001	1.92	1.53–2.42	<0.001
Gender
Male	1.00
Female	0.95	0.78–1.16	0.624
Grade
1	1.00			1.00
2	1.16	0.63–2.12	0.632	1.36	0.70–2.65	0.367
3	1.70	0.91–3.17	0.093	1.82	0.91–3.62	0.09
4	2.18	1.03–4.61	0.042	2.86	1.24–6.59	0.014
Dukes classification
A	1.00			1.00
B	2.09	1.25–3.49	0.005	2.10	1.17–3.74	0.012
C	4.91	2.99–8.08	<0.001	5.43	3.10–9.53	<0.001
D	21.9	13.4–35.9	<0.001	29.4	16.8–51.7	<0.001
Side
Right	1.00
Left	1.01	0.80–1.27	0.963
Location
Colon	1.00
Rectum	1.07	0.88–1.31	0.496
TKTL-1 expression
Low	1.00
High	1.25	1.01–1.56	0.042

Discussion

Transketolase is a key enzyme for the non-oxidative part of the pentose phosphate pathway in glucose metabolism in cancer cells.² We show here high expression of transketolase-like protein 1 (TKTL1), one of the members of the transketolase family, to predict worse survival in colorectal cancer patients, of being associated with Dukes stage B to D, non-mucinous adenocarcinoma, and left-sided disease.

To study glucose metabolism in cancer cells we used immunohistochemistry of the transketolase isoform TKTL1, upregulated in malignancies both at the mRNA and the protein level.^17-19 TKTL1 expression is not a direct measure of glucose metabolism, but mirrors the metabolic activity in cancer cells and works as a tool for estimating glucose metabolism in cancer cells at the time of surgery. TKTL1 tissue expression associates with poor patient survival in various types of cancer.

The few previous studies on tissue expression of TKTL1 in colorectal cancer have been based on rather small patient series. The results reported are, however, promising, and we decided to validate them in our large series of both colon and rectum carcinomas with long follow-up. Schwaab et al. used quantitative PCR to study 33 locally advanced rectal carcinomas receiving neoadjuvant chemotherapy.¹¹ Patients with high expression of TKTL1 had significantly shorter 3-year disease-free survival (DFS) than did those with low expression (34% vs. 87%, P = 0.017). In one series of 70 colon cancers, expression of TKTL1 correlated with tumor invasiveness and poor patient survival.⁴ In another, with 63 patients with colorectal carcinoma, although TKTL1 expression correlated with local tumor progression and regional lymph node metastasis, TKTL1 expression in primary tumors decreased when distant metastasis occured.²⁰ Two alternative reasons for this development were presented, the first based on TKTL1 as enabling metastasis formation, but not being necessary after metastasis occurs. The other is based on a Darwinian assumption of only TKTL1-positive tumors as being able to form metastasis. Both these are hypotheses only. No one has been able to confirm why the decrease in primary tumor TKTL1 levels happens. In our study, high expression of TKTL1 correlated with Dukes stage B-D (P = 0.002), but we failed to see any decrease in primary tumor TKTL1 in Dukes D patients alone. These differences could be due to differing scoring methods. We used manual scoring by 2 researchers, whereas the Diaz-Moralli et al. study used computerized technology. In our study immunohistochemical expression of TKTL1 correlated with poor survival in both colon and rectal carcinomas.

TKTL1-protein expression in colorectal cancer cell lines, when studied with real-time PCR, Western blotting, and immunohistochemistry, was upregulated in about half of colorectal carcinomas but not in healthy tissue, and of 30 patients, 14 showed positive TKTL1 immunostaining.¹⁸ The latter figure is much lower than in our study where 87.5% of tumors stained for TKTL1. Both studies had similar protocols for producing TMA. The difference in results may in part be due to different scoring protocols, as the other study used both staining intensity and percentage of positive tumor cells according to immunoreactive score. Differences were also possible in ischemic time from removal of the specimen to formalin fixation. In our hospital we have for many years paid attention to this part of the process. With the human colon cancer cell line LoVo, it appeared by PCR that mRNA expression of TKTL1 is upregulated.¹⁷ Moreover, when these cells were treated with anti-TKTL1 siRNA, the cell-proliferation rate decreased. Unfortunately we lack data on the mRNA expression in our series.

One of the major differences between normal and cancer cells is this difference in glucose metabolism. The ability of cancer cells to enhance glycolysis is thought to associate with cell proliferation and survival. This specific property of cancer cells offers a possibility to interfere in order to find new treatment methods. As an enzyme, transketolase constitutes a potential target for future anti-cancer therapy. Inhibiting TKTL1 in order might hinder cancer-cells’ growth by limiting their energy sources. Thus far, clinical trials have focused on restricting the glucoseintake of cancer patients. In a study of 20 patients with recurrent glioblastoma, they were each allowed 60 g of carbohydrates per day, and had their urine ketones monitored with urine test sticks. Stable ketosis meant that 2 urine samples per week showed ketones. In patients with stable ketosis - although without any significant effect on progression-free survival (PFS) - a trend for longer PFS was evident, compared to the survival of those failing to reach stable ketosis.²¹ Research on dietary intake of glucose in cancer patients is based on the knowledge of increased glucose metabolism in cancer cells. The theory is that when limiting dietary intake of glucose, less glucose would be available from which cancerous tissue could produce energy from.

The strength of our study is a large, well-characterized patient series of colorectal cancers with long follow-up time and reliable survival data. This material enables subgroup analyses. Although tissue microarray represents only a small part of the tumor, it has proven as good as, or even better than whole sections for analysis of a large patient series. ^22-25

There is no absolute truth regarding scoring of immunohistochemical stainings of prognostic biomarkers. In this study we decided to use the highest score out of 3 spots from each patient for further analysis. We find this appropriate, since TKTL1 expression is known from other studies to correlate with outcome. Therefore, we assumed that the focus with the highest expression represents an aggressive tumor population, that most likely contribute to invasion and metastasis.

In conclusion, our results confirm previous smaller studies’ findings that colorectal cancer patients with high TKTL1 expression have a poor prognosis. Future studies should focus on stage II (Dukes B) colorectal cancer patients of whom 20% develop recurrent disease.¹⁶ In any country these patients will receive adjuvant therapy, although 80% would be cured by surgery alone. It would be important to identify those patients at risk. TKTL1 is thus one candidate marker for risk identification.

Materials and methods

Patients and tissues

This study comprised 840 colorectal cancer patients surgically treated in the Department of Surgery at Helsinki University Hospital, 1989-2000. Of these, 466 (55.5%) were men, and 480 (57.1%) were over 65. The tumor was located in the colon in 429 patients (51.1%), and in the rectum in 411 (48.9%). According to the modified Dukes classification, 125 patients (14.9%) were classified as Dukes A, 294 (35.0%) as Dukes B, 231 (27.5%) as Dukes C, and 190 as Dukes D (22.6%). Disease-specific survival (DSS) at 5 y for all patients was 58.9% (95% CI 55.4-62.4). The median follow-up time was 5.1 y (0–28.0).

Surgical tumor samples were fixed in formalin for more than 24 hours, embedded in paraffin, and stored in the archives of the Department of Pathology. All specimens underwent re-evaluation by an experienced pathologist. Thereafter, the tumor tissue microarray (TMA) blocks constructed included all colorectal cancer tumor samples.^22-24 We collected 3 1.0-mm diameter punches of tissue from representative areas of each of these tumor blocks.

Immunohistochemistry

The TMA blocks were cut into 0.4-mm-thick sections, fixed on slides, and dried for 12 to 24 hours at 37°C. Sections were subsequently deparaffinized in xylene and rehydrated through graded ethanol and distilled water. For antigen retrieval, the slides were treated with Tris-HCl (pH 8.5) in a PreTreatment module (LabVision Corp.) for 20 minutes at 98°C. Staining of sections took place in an Autostainer 480 (LabVision) with an anti-human TKTL1 antibody (Rida Pentocheck IHC, Clone JFC12T10, R-Biopharm AG) diluted 1:200 with Dako REAL Antibody Diluent S2022 (Dako). The primary antibody was kept on glasses over night (O/N) followed by a 30-minute incubation with the secondary peroxide-conjugated rabbit/mouse ENV (K5007) Dako REAL Envision/HRP antibody (Dako). The slides were finally visualized with Dako REAL DAB+Chromogen kept on glasses for 10 minutes. Between each step of the staining procedure, slides were washed with PBS-0.04% Tween20. The slides were counterstained with Meyer's hematoxylin, washed in tap water for 10 minutes, and finally mounted in aqueous mounting medium (Aquamount, BHD). We used a gastric and a colon cancer specimen known to be positive for TKTL1 as positive controls in each staining. Ki67 in this material has been stained in an earlier study using a mouse monoclonal antibody (clone MIB-1, Dako) with a 1:100 dilution.²⁶

Scoring

The samples were scored independently by K.A. and J.H. Cytoplasmic positivity was scored on a 4-grade scale, where absence of staining was scored as 0, mild staining as 1, moderate staining as 2, and strong staining as 3. The highest score out of 3 was chosen to represent the tumor, as high TKTL1 has been shown to associate with poor survival. Samples with no tumor tissue or with too few cells for adequate judging were excluded. Samples receiving different scores from the 2 researchers were discussed to reach consensus.

Statistical analysis

Associations between TKTL1 and clinicopathologic variables were assessed by the chi-square test. Disease-specific overall survival calculation was from date of surgery to date of death from colorectal cancer or until end of follow-up. Survival curves were constructed according to the Kaplan-Meier method and compared with the logrank test. For univariable and multivariable survival analysis, the Cox proportional hazard model had the following covariates entered: age, gender, grade, Dukes classification, histologic type, side (right vs. left), tumor location (colon vs. rectum), and TKTL1 expression. Dukes’ classification, grade, histologic type, and expression of TKTL1 were entered as categorical covariates. A p-value of <0 .05 was considered statistically significant. All statistical analyses were performed with IBM SPSS Statistics version 20.0 for Mac (IBM Corporation).

The Surgical Ethics Committee of Helsinki University Hospital (Dnro HUS 226/E6/06, extension TMK02 §66 17.4.2013) and the National Supervisory Authority of Welfare and Health (Valvira Dnro 10041/06.01.03.01/2012) approved the study.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Päivi Peltokangas and Elina Aspiala for their excellent technical assistance. This work was financially supported by Finska Läkaresällskapet, the Sigrid Jusélius Foundation, and Medicinska Understödsföreningen Liv och Hälsa.

Funding

This work was financially supported by Finska Läkaresällskapet, the Sigrid Jusélius Foundation, and Medicinska Understödsföreningen Livoch Hälsa.