FOXP3+ cell density in lymphoid follicles from histologically normal mucosa is a strong prognostic factor in early stage colon cancer
- First Online:
- Cite this article as:
- Salama, P., Stewart, C., Forrest, C. et al. Cancer Immunol Immunother (2012) 61: 1183. doi:10.1007/s00262-011-1191-3
- 244 Views
There are few clearly established prognostic factors available to guide the use of adjuvant chemotherapy in early stage colon cancer patients. Some of the most promising candidates include the invasion of extramural blood vessels by tumour cells and the densities of FOXP3+ T regulatory cells (Tregs) in tumour and adjacent normal colonic mucosal tissue. The aim of our study was to evaluate the prognostic significance of these markers in AJCC stage II colon cancer, with particular reference to lymphoid follicles in the mucosa. Histopathological review for the presence of vascular and serosal invasion was conducted on a series of 165 stage II colon cancers treated by surgery alone. Immunohistochemical staining for FOXP3 was performed on tumour tissue and histologically normal colonic mucosa from the surgical margin. Image analysis software was used to evaluate the density of FOXP3+ cells in the tumour core, invading margin and lymphoid follicles from the colonic mucosa. For survival analysis, cases were classified into high- or low-density of FOXP3+ cells according to the median value. The mean density of FOXP3+ Tregs in lymphoid follicles was twofold and fivefold higher than in the invading margin and tumour core, respectively. Multivariate analysis identified extramural vascular invasion (HR, 2.47; 95% CI: 1.00–6.07; P = 0.05) and high FOXP3+ cell density in lymphoid follicles (HR, 4.22; 95% CI: 1.49–11.91; P = 0.007) as independent factors for worse survival, whereas a high frequency of lymphoid follicles in histologically normal colonic mucosa was associated with better survival (HR, 0.31; 95% CI: 0.12–0.79; P = 0.014). Our data suggest that host factors related to the immune system have major prognostic significance in early stage colon cancer. The density of FOXP3+ cells within lymphoid follicles and the frequency of these structures in normal colonic mucosa represent novel and independent prognostic factors.
KeywordsFOXP3T regulatory cellsPrognostic markerLymphoid follicleTumour-infiltrating lymphocytesColorectal cancer
The prognosis of colorectal cancer (CRC) has traditionally been estimated using the tumour-node-metastasis (TNM) staging system . The additional histopathological features of tumour cell invasion into extramural vascular and perineural spaces have also proven to be strong and independent risk factors for poor outcome [2–4]. The need for robust and accurate prognostic indicators is especially important for AJCC stage II (T3 or T4, N0, M0), or node-negative colon cancer patients, comprising approximately one-third of all newly diagnosed cases. Better prognostication would allow patients with more aggressive tumours to be considered for adjuvant chemotherapy. The 5-year disease-free survival of stage II colon cancer patients is approximately 70–80% . This varies markedly, however, depending upon the presence or absence of serosal invasion (T3/T4) and the presence of extramural vascular invasion (EMVI) and perineural invasion by tumour cells [2, 4–8].
In addition to TNM staging and vascular/perineural tumour invasion, the presence of a dense lymphocytic infiltrate has consistently been shown to have prognostic value in colon cancer [9–13]. Indeed, some workers have proposed that the density of CD3+ tumour-infiltrating lymphocytes (TILs) may be a more accurate predictor of outcome for CRC than the TNM system [14, 15]. However, a subsequent study found that the prognostic value of TILs was restricted to node-negative cancer , thus ruling out replacement of a TNM-based system with one based upon the anti-tumour immune reaction. Furthermore, these studies did not review the original pathology slides for assessment of important, standard histological markers such as EMVI that may be under-reported. It therefore remains to be established whether the immune response has independent prognostic value in the context of accurate reporting of nodal involvement and of vascular and serosal invasion.
Regulatory T cells (Tregs) are thought to play a major role in cancer through the suppression of anti-tumour immune responses . The transcription factor forkhead box P3 (FOXP3) is a specific nuclear marker for Tregs that allows these cells to be distinguished from other T-cell types. We recently reported that the density of tumour-infiltrating FOXP3+ Tregs, together with vascular and perineural invasion, was an independent prognostic factor in a study of 381 stage II CRCs . In contrast to other cancer types in which high Treg density is associated with poor prognosis, several groups have subsequently confirmed our initial observation of an association with favourable outcome for CRC [19–21]. These results also concur with an earlier report that Treg infiltration was significantly higher in CRC with limited disease stage (UICC I and II) compared with those with more advanced stage (UICC III and IV) . The paradoxical findings for CRC may be due to the suppression by Tregs of a tumour-promoting, inflammatory immune response generated by translocation of bacteria across the mucosal barrier [22, 23].
An additional observation from our earlier study was that high FOXP3+ Treg density in histologically normal colonic mucosa from the surgical margin was associated with poor prognosis . This original finding has yet to be validated in an independent patient cohort. Our earlier study used tissue microarrays and histological information obtained from the initial pathology report. The use of full-face sections would yield a greater area of tumour for analysis and would allow investigation of whether FOXP3+ Tregs at the invasive margin have greater prognostic significance than those within the tumour core. The aim of the present work was therefore to investigate the prognostic significance of FOXP3+ cell densities in neoplastic and normal colonic mucosa from an independent series of stage II colon cancers that were carefully reviewed for the presence of EMVI and serosal invasion. Full-face tissue sections, which are larger and therefore more representative than the cores used in tissue microarrays, were analysed to determine FOXP3+ Treg density in tumour tissue and in histologically normal colonic mucosa from the surgical margin.
Patients with stage II colon cancer (n = 165) who underwent curative resection were identified from a prospectively maintained clinical database and corresponding pathology records. Exclusion criteria were positive surgical margins and the use of adjuvant chemotherapy. Surgery was performed by four specialist colorectal surgeons in a standardized manner to ensure adequate resection margins and lymph node harvest. There were no significant differences between surgeons for either lymph node harvest (mean = 16.8, SD = 7.7) or patient survival. Tumour site was classified as proximal or distal according to location relative to the splenic flexure. Information on patient demographics and tumour features were obtained from the pathology report. Cancer-specific survival information was obtained from the Cancer Registry of Western Australia and from medical records. The median length of follow-up time was 72 months. At the end of the study period, 27 patients (16%) had died from recurrence of their cancer and a further 37 patients (22%) had died from other causes. Ethics approval for the project was obtained from the Human Research Ethics Committees of Fremantle Hospital and St John of God Hospital.
Once cases were identified, the original H&E slides were retrieved and reviewed by a pathologist (CS) for the presence of EMVI as described previously . The pathologist was blinded to the original pathology report. Formalin-fixed, paraffin embedded tissue blocks that optimally demonstrated the invasive tumour margin and histologically normal colonic mucosa from the surgical margin were retrieved for immunohistochemical analysis of FOXP3+ cell density.
Immunohistochemical staining for FOXP3
Sections of 5 ìm thickness were cut from the paraffin embedded tissue blocks, mounted on sialinated slides and subsequently dewaxed and rehydrated using xylene and graded alcohol washes. Antigen retrieval was performed at 121°C for 6 min (DakoCytomation pressure cooker; Dako, Copenhagen, Denmark) with slides placed in Dako target retrieval solution. Endogenous peroxidase activity was blocked (Real Peroxidase Block; Dako) followed by Tris-buffered saline wash. Protein block solution was applied for 10 min to stop non-specific antibody binding (Dako). After incubation for 60 min with primary antibody (FOXP3, Abcam, ab20034, 1/100 dilution), slides were washed in two changes of TBS before incubation with labelled polymer horseradish peroxidase rabbit/mouse antibody for 30 min (Envision Plus Detection System; Dako). Sections were subsequently incubated with Dako-Chromogen solution and washed in deionized water. Counterstaining was performed with Mayer’s haematoxylin and sections then dehydrated through ascending alcohols to xylene and mounted.
Quantification of Treg cell density
Following immunohistochemical staining for FOXP3, slides were scanned with a high-resolution scanner (ScanScope XT; Aperio) at 40× magnification. Image analysis software (Spectrum v10) was used to calculate the density of stained cells (cells per square millimetre) as previously described . The images were examined by one observer (PS) who was blinded to the clinicopathological data and were annotated to ensure that only normal colonic epithelium or viable tumour tissue was included in the area of analysis. Initial attempts to analyse the whole tumour section were not feasible due to multiple software failures resulting from the large size of the images. The density of FOXP3+ lymphocytes was measured at the tumour core and at the invasive margin. The tumour core was defined as solid areas of tumour excluding areas of necrosis, stroma or advancing margin. Using digital image analysis, the number of FOXP3+ lymphocytes was evaluated in five representative areas (each 1 mm2) and an average density was obtained. The invasive margin was defined as the most advanced front of the tumour beyond the muscularis propria and can be considered as the interface between tumour and surrounding tissue. Five representative areas (each 0.25 mm2) were used to calculate the average density of FOXP3+ lymphocytes at the invasive margin. Results for FOXP3+ Treg density were exported into an Excel file and matched to corresponding clinicopathologic data for each case.
Assessment of normal colonic mucosa from the surgical margin
Normal colonic mucosa at the surgical margin was assessed for the total number of lymphoid follicles per centimetre. For proximal colon tumours, the normal sample was always from the distal margin, while for distal tumours, it was from either the proximal or distal ends of the resection. The length of muscularis mucosa was measured with a pen tool, and image analysis software was used to calculate the number of lymphoid follicles per centimetre of normal colonic mucosa. The density of FOXP3+ Tregs within each lymphoid follicle present in or just below the mucosa was evaluated by digital image analysis in the same way as for the tumour sections. The perimeter of individual lymphoid follicles was annotated to determine their area in mm2, and the number of FOXP3+ Tregs within that area was determined using an algorithm. In cases where multiple follicles were present, the average FOXP3+ cell density was determined.
Statistical analysis for cancer-specific survival was performed using Cox proportional hazards regression modelling (STATA version 11 statistical package, STATA Corp, College station, TX). The parameters of FOXP3+ cell density and number of lymphoid follicles per centimetre of mucosal length were classified as high or low in relation to the median value. Logistic regression was used to explore associations between clinicopathological variables and the density of FOXP3+ cells in tumour and normal tissues. For the multivariate analysis, all variables with a P value of <0.1 in univariate analysis were initially included. Non-significant variables were removed sequentially. Kaplan–Meier survival curves with log-rank values were generated for selected variables.
Clinical and histopathological features of 165 stage II colon cancers
Age [years, mean (SD)]
FOXP3+ Treg density (cells/mm2) in tumour tissue and in lymphoid follicles from histologically normal colonic mucosa at the surgical margin
Tumour core (145)
Tumour invasive margin (133)
Normal mucosa lymphoid follicle (137)
In histologically normal colonic mucosa from the surgical margin, the highest density of FOXP3+ Tregs was found in lymphoid follicles and particularly in the mantle zone surrounding the germinal centres (Fig. 1). The FOXP3+ cell density in lymphoid follicles was twofold higher than at the invasive tumour margin (Table 2; P < 0.0001, Wilcoxon rank-sum test). No correlations were observed between FOXP3+ cell densities in the lymphoid follicles and tumour tissue from the same patient, nor with age, gender, tumour site, EMVI or serosal invasion (results not shown). No attempt was made to measure FOXP3+ cell density in the normal mucosal area adjacent to the lymphoid follicles.
The median length of normal colonic mucosa assessed for the presence of lymphoid follicles in each patient was 4.61 cm (mean ± SD, 5.16 ± 2.59). The median frequency of lymphoid follicles from 138 cases for which there was suitable histological material for analysis was 1.12 follicles per cm (mean ± SD, 1.35 ± 1.02; range, 0–5.88). Higher frequencies were found in patients with distal tumours (mean ± SD, 1.67 ± 1.11) compared with those with proximal tumours (1.07 ± 0.87; P = 0.0004, Wilcoxon rank-sum test). No significant associations were found with FOXP3+ cell density or with any other clinical or histopathological features.
Univariate survival analysis for clinicopathological features and FOXP3+ Treg density in stage II colon cancer
Sex (male vs female)
Tumour site (proximal vs distal)
EMVI (yes vs no)
Serosal invasion (yes vs no)
Perforation (yes vs no)
Normal mucosa LF density (high vs low)*
FOXP3+ cells: tumour core (high vs low)
FOXP3+ cells: tumour IM (high vs low)
FOXP3+ cells: normal mucosa LF (high vs low)
Multivariate analysis for indicators of cancer-specific survival in stage II colon cancer
EMVI (yes vs no)
Normal mucosa LF frequency (high vs low)
FOXP3+ cell density in LF (high vs low)
Prognostic factors in colon cancer are of greatest clinical relevance for AJCC/UICC stage II disease (T3 or T4, N0, M0). The identification, validation and routine application of robust prognostic markers would allow node-negative patients with poorer survival prospects to be considered for adjuvant chemotherapy, while sparing many others the toxicity and expense of such treatment. This study found a strong association between EMVI and worse outcome (Table 4; Fig. 2), thus confirming several previous reports showing independent prognostic value for this feature in CRC [2, 4, 5, 7, 8, 24]. Although not reaching statistical significance, the present results also confirm the prognostic value of tumour-infiltrating FOXP3+ Tregs [18–21, 25]. Furthermore, the results support a recent finding by our group that high FOXP3+ Treg density in the normal colonic mucosa was associated with poor survival . This previous observation has now been confirmed and extended in a separate patient cohort where high FOXP3+ cell density within mucosal lymphoid follicles was found to be a strong and independent factor for unfavourable outcome (Table 4). Moreover, we report for the first time that the frequency of these lymphoid follicles in the colonic mucosa also showed prognostic value.
The need for careful evaluation of EMVI has been highlighted by several groups over the past decade [2–4, 6, 7, 24]. Approximately 10–15% of CRCs were reported to show EMVI in routine practice in the UK  and Australia . However, it has been argued that proper specimen preparation and more thorough evaluation by pathologists increases this frequency to around 25–30% [2, 3]. All cases in the present study were reviewed, resulting in a frequency of 25% for EMVI. The hazard ratio (HR) associated with EMVI was 2.46 in a multivariate analysis that included the FOXP3+ marker (Table 4). This result compares with a HR of 2.16 in our previous study of 381 stage II colon cancers that also included FOXP3+ cell density but in which the pathology was not reviewed . An earlier population-based study by our group of 1,306 stage II colon cancer patients found a HR of 1.63 for vascular invasion; however, again the pathology was not reviewed and the reported frequency was only 12% . This latter result is similar to another study of 362 node-negative CRCs in which the HR for venous invasion was reported to be 1.96, but its frequency was just 13% . The importance of careful histological review was recently highlighted in a study of 381 node-negative CRCs that showed the HR associated with venous invasion (23% frequency) was 4.45 following review, but only 1.05 based on routine reporting . Taken together, the above findings demonstrate that EMVI, providing it is carefully evaluated, will be a key factor in future algorithms for the prognostic stratification of stage II colon cancer patients.
A large volume of literature clearly demonstrates that a strong TILs reaction is associated with favourable prognosis in CRC [9–15]. This is likely to be due to the suppression of early metastatic invasion by a local immune response involving activated T cells [14, 15]. Importantly, the good prognosis associated with a high density of CD3+ TILs appears to be confined to patients with node-negative disease . We have shown that the density of FOXP3+ Tregs in CRC shows stronger prognostic significance than the density of CD8+ and CD45RO+ cell markers of the cytotoxic immune response . In most cancer types, Tregs suppress anti-tumour immune responses, and high densities of these cells are associated with worse patient outcome . However, several studies have now shown that a high density of FOXP3+ Tregs in CRC correlates with better survival [18–21, 25]. Ladoire et al.  have argued that the paradoxical result observed for CRC could be due to Treg-mediated suppression of a tumour-promoting, inflammatory immune response that is generated by translocation of bacteria across the mucosal barrier. Most cancers occur in relatively sterile environments, whereas colon cancers occur in a highly contaminated environment in which the immune system is geared towards tolerance of the bacteria.
The good prognosis associated with a high density of FOXP3+ Tregs in the tumour core (HR 0.73) and invasive margin (HR 0.63) did not reach statistical significance in the present study of 165 stage II colon cancers (Table 3). However, these HR values were similar to those reported in our previous study of 967 stage II and III CRCs (HR, 0.78)  and to other large studies of 613 mismatch repair-proficient and 223 mismatch repair-deficient CRC (HR, 0.70; and HR, 0.63, respectively)  and of 768 stage I–IV CRCs (HR, 0.48) . It is likely that the low number of cancer-related events (n = 26) in this good prognosis cohort prevented the density of tumour-infiltrating FOXP3+ cells from reaching significance as a prognostic marker for favourable outcome. Further work is required to determine whether the density of tumour-infiltrating FOXP3+ cells is an independent prognostic marker in stage II colon cancers in which EMVI has been carefully evaluated.
We previously reported that high FOXP3+ cell density in the normal colonic mucosa from the surgical margin was associated with worse patient outcome . In the present work, we have confirmed and extended this original finding in a separate cohort by showing that high FOXP3+ cell density in the lymphoid follicles was a strong (HR 4.22) and independent factor for poor survival (Table 4). An unexpected and novel observation from the current study was that the frequency of lymphoid follicles in the colonic mucosa was also an independent prognostic factor. The total number of lymphoid follicles in the human large bowel has been estimated from autopsy specimens to range from 12,000 to 18,000 . Their number and diameter have been shown to increase in inflammatory conditions , while some workers have proposed that lymphoid follicles are intimately involved in mucosal regeneration as well as in immune surveillance . The density of intratumoural lymphoid structures has been associated with improved survival in non-small cell lung cancer . Another study has reported that the incidence of lymphoid follicles associated with early colorectal neoplasms varies according to patient gender and to tumour site and histology .
The present study is the first to investigate the prognostic significance of FOXP3+ Treg density in lymphoid follicles from normal colonic mucosa (Table 4). The worse outcome of patients with high Treg density is an intriguing finding, and we hypothesize that this could be an indicator of increased systemic susceptibility to metastasis. It will be interesting to determine whether other immune cell subtypes present in follicles also have prognostic significance and whether the density of FOXP3+ cells in follicles correlates with their level in circulation. The predominant localization of FOXP3+ Tregs to the mantle zone (Fig. 1a, b) is similar to the report by Lim et al.  in tonsil sections. Although not investigated here, lymphoid follicles with small germinal centres (Treg poor) would thus be expected to show higher FOXP3+ cell densities compared with those with relatively larger centres. The apparently worse outcome associated with high Treg density in mucosal lymphoid follicles (Table 4) could therefore be a function of impaired B-cell reaction, as observed by a smaller germinal centre response, rather than a direct effect of Tregs.
To our knowledge, the present study is also the first to investigate the prognostic significance of lymphoid follicle frequency in the adjacent non-tumour tissue of any cancer type. The observation of better outcome for patients with a high frequency of lymphoid follicles (HR, 0.31) could be a potentially important finding if confirmed by further studies. This factor is relatively straightforward to quantify and provides prognostic information that appears to be independent of both EMVI and FOXP3+ cell density (Table 4). In exploratory subgroup analysis, the lymphoid follicle frequency appeared to have a stronger prognostic significance for patients with proximal tumours (HR = 0.13; P = 0.051; 95% CI: 0.02–1.00) compared with those with distal tumours (HR = 0.64; P = 0.488; 95% CI: 0.18–2.27). It is currently unclear why this factor has prognostic significance; however, we suspect that it could reflect the status of the alimentary or systemic immune system, including its ability to respond to neoplasia.
Based on the present work and on earlier studies discussed above, there is now abundant evidence that EMVI and TILs can provide clinically useful information for the prognostic stratification of stage II colon cancer patients. For EMVI, the need for high-quality pathology reporting and optimal specimen preparation has already been emphasized [2, 3, 24]. For TILs, further work is required to determine the immune parameter(s) that provide the strongest and most robust prognostic information. The current study and our previous investigation  indicate that histologically normal colonic mucosa should not be overlooked as a potential source of clinically relevant markers. In particular, the density of FOXP3+ cells within lymphoid follicles and the frequency of these in the mucosa represent novel and independent prognostic factors. These should now be validated in additional, large cohorts of stage II colon cancers in which a minimum number of lymph nodes have been examined, EMVI has been carefully reviewed and patients have not received adjuvant chemotherapy.
Paul Salama was supported by a Sir Walter Gibbon Postgraduate Research Award from the University of Western Australia and by a grant from the Colorectal Surgical Society of Australia and New Zealand Foundation. The authors acknowledge the facilities and scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy, Characterisation & Analysis, University of Western Australia. This facility is funded by the University of Western Australia and state and Commonwealth Governments. The authors gratefully acknowledge the assistance of Dr Tim Threlfall from the Western Australian Cancer Registry and Ms Lisa Spalding for technical assistance with immunohistochemistry. This work was funded by the Cancer Council of Western Australia and St John of God Hospital, Subiaco.
Conflict of interest
The authors declare that they have no conflict of interest.