• colorectal cancer;
  • surveillance;
  • microsatellite instability;
  • DNA ploidy;
  • recurrence;
  • locoregional;
  • metastasis;
  • lung;
  • liver;
  • survival


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References


Appropriate stratification tools for targeted surveillance after resection for colorectal cancer (CRC) are lacking. The objective of the current study was to investigate the effect of microsatellite instability (MSI) and DNA ploidy on surveillance after surgery.


The authors evaluated 186 consecutive, population-based patients with stage I through III CRC who underwent surgery with curative intent and who entered a systematic surveillance program. MSI was analyzed with polymerase chain reaction for 5 known quasimonomorphic markers (BAT-26, BAT-25, NR-21, NR-24, and NR-27), and DNA ploidy was analyzed with automated cytometry. Recurrence, recurrence-free survival (RFS), and disease-specific survival (DSS) were evaluated by univariate and multivariate statistical tests.


Patients with MSI (20%) were significantly younger than patients without MSI (median age, 61 years vs 67 years; P = .016). Proximal location (adjusted odds ratio [AOR], 5.4; 95% confidence interval [95% CI], 2.1-14.1 [P = .001]), large tumor size (≥5 cm: AOR, 3.5; 95% CI, 1.3–9.6 [P = .015]), and poor tumor differentiation (AOR, 6.6; 95% CI, 2–21.8 [P = .002]) were associated with MSI. MSI conveyed an increased risk for locoregional recurrence (OR, 2.9; 95% CI, 1.2–7 [P = .016]), with a trend toward a shorter time to recurrence (P = .060). Neither MSI status nor DNA ploidy predicted distant metastasis, RFS, or DSS. Lymph node status was the best predictor of distant spread (AOR, 3.9; 95% CI, 2–7.9 [P < .001]) and DSS (hazard ratio, 4.9; 95% CI, 2.6–9 [P < .001]).


Patients who had microsatellite instable tumors were at increased risk for locoregional recurrence, whereas lymph node status was the best predictor of distant metastasis. Clinical surveillance and choice of modality (ie, endoscopy vs radiologic imaging) may be improved when patients are stratified according to these cancer features. Cancer 2009. © 2009 American Cancer Society.

Colorectal cancer (CRC) is a major contributor to cancer-related deaths in the Western world and poses a considerable global health burden with nearly 1 million new cases reported annually.1 It has been established that CRC develops through 3 major genetic pathways, incorporating gross chromosomal instability (CIN), epigenetic changes, and microsatellite instability (MSI).2 Surgery is the mainstay therapy for cure, and survival is among patients with CRC dictated by the stage of disease at presentation. Although adjuvant chemotherapy is administered to patients with stage III disease (lymph node metastasis present or pathologic lymph node-positive [pN+]), prognostic and predictive features may overlap among stages. Furthermore, recurrent and metastatic disease may occur (stage-dependent) in from 30% to 40% of all patients who undergo ‘curative’ surgery. Thus, surveillance programs have been developed to detect asymptomatic locoregional recurrence or distant spread at an early stage when secondary curative surgery can be offered.

Although a current meta-analysis of available data suggested a survival benefit for patients undergoing intense follow-up after surgery,3 the optimal mode of follow-up (clinical examination, endoscopy, blood tests, and radiologic imaging) remains to be defined, as evident from the wide number of societal and national surveillance strategies offered.4–6 Consequently, heterogeneity is reported concerning compliance, effectiveness, costs, and survival benefits among studies.4, 6–9

The only tumor marker in widespread clinical use for CRC surveillance is the carcinoembryonic antigen (CEA). The large number of tests required for serial serum measurement of CEA to detect 1 (asymptomatic) curable recurrence makes this test costly from an overall cost-benefit perspective.6 In addition, the diagnostic accuracy of CEA makes correct interpretation a clinical challenge.10 Thus, new and better risk stratification and surveillance markers are needed to better tailor the postsurgery surveillance of patients with CRC.

The increased understanding of the molecular biology in colorectal carcinogenesis has revealed several promising diagnostic, prognostic, and predictive biomarkers derived from the major pathways involved.11–13 CIN is associated with aneuploid tumors; thus, ploidy serves as a surrogate marker for CIN in CRC.11, 12 Most previous studies on MSI status in patients with CRC have focused on survival and response to chemotherapy, for which recent data continue to produce equivocal results.14–19 Conversely, few, if any, studies have investigated the effect of MSI status in CRC on recurrence patterns and survival among patients who were included in a strict surveillance program. Thus, in this study, we investigated the role of MSI and CIN on the influence on recurrence (locoregional or distant) and disease-specific survival (DSS) in a defined patient cohort that underwent systematic surveillance after curative resection of CRC.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References

Study Cohort

The population under study has been described thoroughly in previous reports.6, 10 Briefly, 314 consecutive patients with stage I through III CRC underwent surgery with curative intent (ie, with negative resection margins [R0]) between 1996 and 1999 at the Stavanger University Hospital, which serves as the only hospital for a defined population of 290,000 inhabitants. The patients under study were from a defined, unselected population (Fig. 1). Consequently, the results obtained should be generally applicable to other Western populations.

thumbnail image

Figure 1. Patient selection and inclusion criteria for the study. MSI indicates microsatellite.

Download figure to PowerPoint

Of the 314 patients who underwent R0 surgical resection, 194 patients were enrolled on a systematic surveillance program in an intent-to-treat basis and according to the guidelines of the Norwegian Gastrointestinal Cancer Group.6 During the study period, patients with rectal cancers underwent total mesorectal excision without routine use of preoperative radiochemotherapy according to national guidelines at the time. Patients with stage III colonic cancers routinely were offered a 5-fluorouracil/leucovorin-based adjuvant chemotherapy regimen according to the national standards during the treatment period. The 120 patients (38%) who did not receive systematic follow-up were significantly older (median age, 81 years) and had more comorbidities (indicated by higher American Society of Anesthesiologists [ASA] scores), which made them unlikely candidates for either extensive chemotherapy or secondary surgery in the case of recurrence at the time when the study was performed.6 Follow-up was continued until death from CRC or from any other cause, and patients who remained alive were censored as of August 1, 2005. Of the previously reported 194 patients, 186 patients (96%) were eligible for the current study (Fig. 1). All cancers were reviewed by a board-certified pathologist, classified according to World Health Organization criteria and staged according to the American Joint Committee on Cancer TNM classification system. All cancers were subject to macroscopic and microscopic systematic and histopathologic investigation, but no additional pathology features, such as venous or lymphatic invasion, tumor lymphocyte infiltrate, or other markers, were investigated at the time of this study.

Any first recurrence of disease was recorded during follow-up and was defined as either locoregional recurrence (any intraluminal or regional peritoneal/lymph node recurrence), or distant metastasis (hepatic, pulmonary, or any other/multiple sites), or both. Proximal sites in the colon were defined as disease locations in the cecum through the left splenic flexure, and sites in the distal colon were defined as disease locations in the descending colon through the rectum.

Microsatellite Instability Analysis

Microsatellite analysis was performed on DNA that was isolated from archived, paraffin-embedded tissue blocks using tissue from the most invasive front of each tumor (reviewed and identified on hematoxylin and eosin-stained slides, avoiding areas with extensive necrosis or normal tissue), using the QIAamp DNA Micro-Kit (QiaGen, Hilden, Germany) and the manufacturer's protocol for DNA-isolation. MSI analysis was performed with 5 markers (Table 1) that have been established as quasimonomorphic and have a low risk for polymorphisms (and, thus, a low rate of false-positive detection of MSI) in the Caucasian population, as described previously.20, 21 These selected markers have a high sensitivity for MSI without the need for matching with patients' normal DNA, as reported by others.22 Polymerase chain reaction (PCR) amplification was performed under standard conditions. Normal DNA for control was extracted from blood from a healthy laboratory technician who had no prior history of cancer or any known hereditary cancer syndrome. The reactions were incubated with a denaturation step at 94°C for 4 minutes followed by 35 cycles of 94°C for 1 minute, 56°C for 2 minutes, 72°C for 2 minutes, and a final extension step at 70°C for 10 minutes. The amplified PCR products were run on an automated 16-capillary electrophoresis DNA sequencer (GeneAnalyzer; 3130XL) and allelic sizes were estimated with GeneMap software. MSI in any marker was observed as a shift in the product sequence. Instability in ≥2 of 5 markers (≥40%) was regarded as high-frequency MSI (MSI-H), in instability 1 of 5 markers was regarded as low-frequency (MSI-L), and instability in no markers was regarded as stable (MSS). In this study, tumors with MSI-L were regarded as MSS, in line with previous reports.20

Table 1. Primers Used for Microsatellite Analysis
Marker*GenePrimer SequenceGene Bank No.Average PCR Product Size, Base Pairs
  • PCR indicates polymerase chain reaction; hMSH2, human mutS homolog 2; F, forward primer sequence; R, reverse primer sequence; SLC7A8, solute carrier family 7, member 8; ZNF-2, zink-finger 2; IAP-1, inhibitor of apoptotis protein 1.

  • *

    The fluorescent markers used were FAM for BAT-26, NR-24 and NR-27, and HEX for BAT-25 and NR-21.

BAT-25c-kitF: 5′-TCG CCT CCA AGA ATG TAA GT-3′L04143124
NR-21SLC7A8F: 5′-GAG TCG CTG GCA CAG TTC TA-3′XM_033393110
NR-24ZNF-2F: 5′-GCT GAA TTT TAC CTC CTG AC-3′X60152125

DNA Ploidy Analysis

Cell suspensions were prepared from archived, paraffin-embedded blocks. Cytospins were Feulgen-stained with pararosanilin under standardized conditions, and cytometric analysis was performed on a fully automated DNA image cytometer (QPath; Leica, Cambridge, UK).23, 24 At least 3000 objects per slide (and up to 5000 objects per slide) were scanned automatically using predefined densitometric and geometric filters. The following histogram descriptors were calculated from each cancer: ploidy (diploid or nondiploid), analysis of S-phase, the percentage of cells in the histogram with a DNA content exceeding 2.5C and 5C (referred to as the ‘c-exceeding rate’ [cER]; or 2.5cER and 5cER), and the Boecking malignancy grade.24 The Multicycle program (Phoenix Flow Systems, San Diego, Calif) was used to calculate the percentage of cells in G0,G1-phase, S-phase, G2M-phase, and above G2M-phase in the cell cycle. Tumors with a DNA index (DI) >0.9 or <1.1 were regarded as diploid, whereas tumors with a DI <0.9 or ≥1.1 were regarded as aneuploid.


All aspects of this study were consented and approved by the Regional Ethics Committee, the Norwegian Social Science Data Service, and the Norwegian Data Inspectorate.

Statistical Analysis

Data were analyzed using SPSS software (version 13 for Macintosh; SPSS Inc., Chicago, Ill). Continuous variables were analyzed by using either the Student t test or the Mann-Whitney U test (for variables with a nonparametric distribution).

Categorical variables were analyzed with the chi-square or the Fisher exact test, as appropriate. The evaluation of features associated with MSI tumors or with developing recurrent disease was performed using 2 × 2 tables for univariate analysis and logistic regression for multivariate analysis to calculate odds ratios (OR) with 95% confidence intervals (95% CIs).

Differences in Kaplan-Meier survival curves were analyzed with the log-rank test for the time to recurrence (overall, and locoregional, and distant recurrence), recurrence-free survival (RFS) (using any recurrence as the endpoint), and DSS (using death from CRC as the endpoint). Multivariate analysis was performed with a Cox proportional hazards model. Age, sex and MSI status were included in the model in addition to variables that had P values <.1 on univariate analysis. All tests were 2-tailed, and statistical significance was set at P < .05.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References

Patient and Tumor Characteristics

Clinicopathologic and demographic data are provided in Table 2. The ratio of men to women was 3:2, and MSI was distributed equally among the sexes. The median age at diagnosis was 67 years (interquartile range [IQR], 59–73 years). Patients with MSI tumors were significantly younger than patients with MSS tumors (median, 61 years vs 68 years, respectively; P = .016). The majority of patients underwent elective surgery for CRC, and 12 patients (7%) underwent surgery in an emergent manner, but no differences were observed in their MSI status, recurrence rates, or survival. The overall median follow-up was 6.2 years (IQR, 4.5–7.3 years) and did not differ significantly between patients with MSI tumors and patients with MSS tumors (P = .859).

Table 2. Tumor Features in Microsatellite Stable and Instable Colorectal Cancers
 No. of Tumors (%)    
Tumor CharacteristicTotalMSSMSIOR [95% CI]P*Adjusted OR [95% CI]P
  • MSS indicates microsatellite stable; MSI, microsatellite instable; OR, odds ratio; 95% CI, 95% confidence interval; pN, pathologic lymph node classification; pT, pathologic tumor classification.

  • *

    A chi-square or Fisher exact was test used when appropriate (with 1 degree of freedom).

  • On logistic regression.

  • Calculated for stage I disease versus stage II and III disease; a chi-square test was used for stage I versus II and III disease (with 1 degree of freedom).

 Rectum65 (35)6141 1 
 Colon121 (65)88335.7 [1.9–17].0011.4 [0.3–6.2].629
 Distal121 (65)11291 1 
 Proximal65 (35)37289.4 [4.1–21.8]<.0015.4 [2.1–14.1].001
Tumor size, mm
 <5095 (55)8871 1 
 ≥5079 (45)54255.8 [2.4–14.4]<.0013.5 [1.3–9.6].015
Lymph node status
 pN0129 (69)101 (78)28 (22)1 
 pN+57 (31)48 (84)9 (16)0.7 [0.3–1.5].351  
Tumor grade
 Well/moderate164 (88)140241 1 
 Poor/mucinous22 (12)9138.4 [3.2–21.9]<.0016.6 [2–21.8].002
 Diploid109 (61)74351   
 Aneuploid69 (39)6900.7 [0.6–0.8]<.001.996
 Low, <5%78 (44)68101 1 
 High, ≥5%100 (56)75252.3 [1.0–5.1].0431.9 [0.7–5.1] 
pT classification
 pT1-T239 (21)3721 1.182
 pT3-T4147 (79)112355.8 [1.3–25.2].0092.3 [0.3–20].451
TNM stage
 I33 (18)3211.0071 
 II96 (52)69279.8 [1.3–74.6] 2 [0.2–17.7].526
 III57 (31)489    

Thirty-seven patients (20%) patients had MSI-H tumors, including 34 (92%) who were positive for MSI in at least 4 of the 5 markers. Five patients (3%) had tumors that were classified as MSI-L and had only 1 marker that was positive for MSI (2 with instability for BAT-26 only and 1 each with instability for BAT-25, NR-21, and NR-24). Typical PCR product sizes and shift in sizes for MSS and MSI tumors were observed for each marker (data not shown).

The patients' tumor characteristics (Table 2) underline many of the specific features associated with MSI in colorectal tumors and serve as a validation to previously reported studies. The table indicates that MSI tumors were observed significantly more often in colonic sites versus rectal sites, demonstrating a proximal predilection for MSI, with 28 of 65 proximal cancers (43%) classified as MSI compared with only 9 of 121 distal cancers (7%). The MSI tumors were larger, more invasive (pathologic tumor classification 3 or 4 [pT3-pT4]), and at a more advanced stage (stages II or III) than their MSS counterparts in addition to having a predominant diploid DNA content and poor/mucinous histopathologic differentiation. No significant difference in lymph node status was observed. Logistic regression analysis adjusted for sex and age retained proximal location, large tumor size (>5 cm), and poor histopathologic differentiation as the strongest tumor features associated with MSI (Table 2).

Overall Recurrence Rates and Time to Disease Recurrence

Seventy patients (38%) developed recurrent disease, and 50 of those patients developed metastasis of the liver (n = 22 patients), lung (n = 13 patients), or various other sites (n = 15 patients; including multiple locations in 6 patients and central nervous system and skeletal locations in 2 patients each). Locoregional recurrence developed in 27 patients (including 7 patients who also had distant metastasis). There were no statistically significant differences in the overall recurrence rate between the sexes, ages (≥67 years vs <67 years), primary tumor location (colon vs rectum), subsite (distal vs proximal), tumor size (≥5 cm vs <5 cm), pT classification (pT1-T2 vs pT3-T4), or grade. The factors that were analyzed for the risk of any recurrence are listed in Table 3. The OR for having any recurrence was significantly greater for patients with stage III tumors, which was explained by the presence of lymph node metastasis with an adjusted OR of 3.4 for patients who had pN+ tumors versus pN0 tumors (95% CI, 1.8–6.5; P < .001).

Table 3. Tumor Characteristics Associated With the Risk of Developing Any (Both Locoregional and Distant) Disease Recurrence
 No. of Tumors (%)  
CharacteristicNo RecurrenceRecurrenceOR [95% CI]P*
  • OR indicates odds ratio; 95% CI, 95% confidence interval; pT, pathologic tumor classification; pN, pathologic lymph node classification; MSS, microsatellite stable; MSI, microsatellite instable.

  • *

    Chi-square test (with 1 degree of freedom).

  • None of the DNA histogram features (2.5C or 5C exceeding rate, Boecking malignancy grade) were informative.

 All patients116 (62)70 (38)
TNM stage
 I/II92 (71)37 (29)1<.001
 III24 (42)33 (58)3.4 [1.8–6.5] 
pT classification
 pT1-T228 (72)11 (28)1.172
 pT3-T488 (60)59 (40)1.7 [0.8–3.7] 
Lymph node status
 pN092 (71)37 (29)1<.001
 pN+24 (42)33 (58)3.4 [1.8–6.5] 
Microsatellite status
 MSS94 (63)55 (37)1.684
 MSI22 (60)15 (40)1.2 [0.6–2.4] 
DNA ploidy
 Diploid68 (62)41 (38)1.875
 Aneuploid44 (64)25 (36)0.9 [0.5–1.8] 
DNA S-phase
 Low, ≤5%48 (62)30 (39)1.736
 High, >5%64 (64)36 (36)0.9 [0.5–1.7] 

Although patients who had positive lymph node status appeared to develop recurrences earlier compared with patients who had pN0 status, there were no significant differences in the time to recurrence (P = .096) in the Kaplan-Meier log-rank tests. However, both poor tumor grade (P = .001) and younger age (≤60 years; P = .036) were associated with a shorter time to overall recurrence.

Locoregional Disease Recurrence

Patients who had tumors with MSI-H had a significantly greater risk for developing locoregional recurrence compared with patients who had MSS tumors (10 of 37 patients [27%] for MSI vs 17 of 149 patients [11%] for MSS; OR, 2.9; 95% CI, 1.2–7.0 [P = .016]). MSI retained its statistically predictive ability when adjusted for both age and sex (adjusted OR, 2.5; 95% CI, 1.0–6.1 [P = .047]), and when adjusted for tumor location (adjusted OR, 3.1; 95% CI, 1.1–8.8 [P = .026]). In addition, a trend toward a shorter time to locoregional recurrence (P = .060) was noted for patients who had MSI tumors (Fig. 2). None of the other features, including DNA ploidy or histogram descriptors, were associated significantly with the risk of locoregional recurrence or with the time to locoregional recurrence. It is noteworthy that pN+ lymph node status did not predict locoregional recurrence (OR, 2.0; 95% CI, 0.9–4.7 [P = .093]).

thumbnail image

Figure 2. The time to locoregional recurrence was analyzed according to microsatellite status. MSI-H indicates high-frequency microsatellite instability; MSS, microsatellite stable.

Download figure to PowerPoint

Distant Metastasis

Fifty patients (27%) developed distant metastasis during surveillance after primary curative surgery for CRC. MSI was not associated with distant metastasis (OR, 0.6; 95% CI, 0.3–1.6 [P = .321]). However, positive lymph node status was associated significantly with the risk of developing distant metastasis (27 of 57 patients [47%] for pN+ vs 23 of 129 patients [18%] for pN0; OR, 4.1; 95% CI, 2.1–8.3 [P < .001]), which remained significant when adjusted for age, sex, and tumor location (adjusted OR, 3.9; 95% CI, 2.0–7.9 [P < .001]). Positive lymph node status also predicted a significantly shorter time to the development of distant metastasis (P = .019) along with low tumor grade (P = .002). The time from surgery to a diagnosis of metastasis did not differ significantly among patient characteristics (age, sex), MSI status, or ploidy status (including DNA histogram descriptors) or for any other tumor characteristics (size, stage, lymph node status). However, poor tumor grade of the primary tumor predicted the development of distant metastasis within a significantly shorter time, and all such patients developed recurrent disease within 3 years after surgery (P = .016; hazards ratio, 3.6; 95% CI, 1.6–8.8).

Recurrence-free Survival

TNM classification and lymph node status were the only significant predictors of RFS. In multivariate Cox proportional hazards analysis, lymph node status (pN0 vs pN1 vs pN2) was the only significant predictor of RFS that was retained in the model. MSI did not predict RFS for all patients (P = .678) or for the select analysis of patients with stage II and III CRC (P = .759).

Disease-specific Survival

At the end of follow-up, 125 patients remained alive (67%), and 44 patients had died of CRC (24%), whereas 17 patients had died from other causes, including 10 deaths from other cancers. Of the patients who had any recurrent disease (n = 70 patients), in total, 26 (37%) remained alive with disease. Neither age nor sex was predictive of DSS. MSI did not predict DSS for all patients with CRC (P = .920) or for the select analysis of patients with stage II and III CRC (P = .722). Significant univariate discriminators for DSS were proximal/distal location, TNM classification (Fig. 3A), pT classification, pN status (Fig. 3B), and tumor grade (Table 4). Lymph node status was the most important prognosticator in the multivariate analysis model for DSS (hazards ratio, 4.9; 95% CI, 2.6–9.0 [P < .001]).

thumbnail image

Figure 3. This Kaplan-Meier survival curve illustrates disease-specific survival according to (A) TNM stage and (B) lymph node status. The corresponding hazards ratios (HR) with P values are presented in Table 4 for TNM stages (A). Lymph node status (B) was retained in the multivariate analysis, with increasing risk noted for each lymph node stage. Values are shown for pathologic N1 (pN1) tumors (HR, 4.3; 95% confidence interval; [95% CI], 2.3–8.3 [P < .001]) and for pN2 tumors (HR, 7.9; 95% CI, 3.2–19.3 [P < .001]) compared with pN0 tumors.

Download figure to PowerPoint

Table 4. Disease-specific Survival With Univariate and Multivariate Analysis*
  Univariate AnalysisMultivariate Analysis
CharacteristicNo. of Events (%)HR [95% CI]Log-Rank PHR [95% CI]P
  • HR indicates hazards ratio; 95% CI, 95% confidence interval; pT, pathologic tumor classification; pN, pathologic lymph node classification; MSI, microsatellite instable; MSI-H, high-frequency microsatellite instability; MSS, microsatellite stable.

  • *

    Numbers (and percentages) may not add up because of rounding or missing values. The reference category for HR is listed first.

  • None of the DNA histogram features (2.5C or 5C exceeding rate, Boecking malignancy grade) were informative.

Primary site 
 Rectal15/65 (23)    
 Colonic29/121 (24)    
Subsite  .042.326
 Distal23/121 (19)    
 Proximal21/65 (32)1.8 [1.0–3.3]   
Tumor size, mm .258.612
 <5020/95 (21)    
 ≥5022/79 (28)    
TNM stage   .087
 I2/33 (6)    
 II14/96 (15)2.5 [0.6–10.9].228  
 III28/57 (49)10.2 [2.4–42.8].002  
pT classification  .029.070
 pT1-T24/39 (10)    
 pT3-T440/147 (27)3.0 [1.1–8.3]   
Lymph node status  <.001 <.001
 pN016/129 (12)  4.9 [2.6–9.0] 
 pN+28/57 (49)4.9 [2.6–9]   
Tumor grade  .003.190
 Well/moderate34/164 (21)    
 Poor/mucinous10/22 (45)2.8 [1.4–5.7]   
MSI status .920.371
 MSI-H9/37 (23)    
 MSS35/149 (23)    
DNA ploidy .888
 Diploid25/109 (23)    
 Aneuploid17/69 (25)    
DNA S-phase, % .436
 ≤516/78 (20)    
 >526/100 (26)    


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References

Most patients with CRC who undergo treatment with curative intent subsequently enter a surveillance program. The primary objective of surveillance is to identify patients with disease recurrence at a resectable stage. However, the identification of local recurrence and metachronous (new) carcinoma also is an important aspect of follow-up; thus, endoscopy plays a central part in several surveillance guidelines.4, 6, 7, 25, 26 Patients who are under observation may be referred for more extensive imaging, either because regular imaging forms part of the surveillance strategy or because tumor recurrence is suggested by the development of new symptoms, or a rise in tumor marker levels, most often the serial assessment of CEA.10 Although evidence points toward a survival benefit for patients who undergo surveillance, the optimal protocol remains to be determined. Moreover, experience in clinical practice may differ from the results obtained in randomized trials.

Stage (ie, stage III, pN) of the primary tumor has been regarded as the strongest prognostic and predictive factor and, thus, guides the current use of adjuvant therapy. However, recurrences occur even in patients who have disease in a favorable stage (ie, stage I and II); thus, the worldwide surge to identify better stratification markers to guide surveillance and risk stratification in patients with CRC continues.

In the current study of patients with CRC from a defined population undergoing systematic surveillance after curative surgery, we observed MSI-H in 20%, confirming many of the distinct features associated with MSI tumors of the colorectum.2, 27 Although MSI tumors have features that generally are associated with a poor prognosis, such as large size, invasive growth pattern, and poorly differentiated tumors, we could not demonstrate any statistically significant differences in disease-free survival or the time to overall recurrence in the current study. Although a previously performed meta-analysis on all studies published up to 2003 stated that there was an overall weak yet significant survival benefit from having MSI,13 our study was in line with more recently obtained results reporting no significant differences in DSS for patients who had the MSI genotype compared with patients without MSI.14, 15, 17–19, 28–30 It is noteworthy that, in recent years, laboratory molecular diagnostics and standardization of MSI testing have improved; however, the determination and interpretation of MSI remains an ongoing clinical challenge.16, 31 Also, the discrepancies in the reported survival results also reflect the heterogeneity of the populations studied; eg, unselected or institutional series of sporadic CRC,14, 32 hereditary cancer (hereditary nonpolyposis CRC) screening,19 analysis of adjuvant chemotherapy trials,15, 17 site- and/or stage-specific analyses (ie, only colonic cancers, only stage II and/or III cancers), as well as in age-related, geographic, and ethnical differences.20, 33

It is also is worth noting that, in the current study, the majority of older patients (aged >75 years) were excluded from the systematic surveillance program and, as such, were not included in the current survival analysis. We previously demonstrated that, in the group aged >75 years, overall survival was poorer, but DSS was comparable to that among patients who had systematic surveillance.6 We speculate that this may point to a beneficial tumor biology in the oldest age group, such as having a large proportion of MSI cancers. It also is worth noting that MSI is an age-related phenomenon: It has a bimodal MSI distribution at the extremes of age, and up to 33% of all CRC in the elderly is positive for MSI.33 MSI also has demonstrated prognostic value in very young CRC patient populations34 and, in several heterogeneous (stage-dependent) analyses, included patients of all ages derived from trials on adjuvant chemotherapy or selected cohorts of patients.14, 15, 17, 18, 28, 32, 35, 36 In the current population-based study on surveillance after surgery, the exclusion of elderly patients may have masked an overall survival benefit with regard to MSI. However, the current results are in line with several other studies, which reported no statistically significant difference in DSS for patients who had MSI tumors, although the results demonstrated nonsignificant trends or only stage- dependent improved overall survival for patients who had MSI tumors.17, 18, 29 Regardless of those reports, and as stated above, considerable comorbidity (reflected by high ASA scores) in the elderly patients who were not eligible for extensive follow-up in this study also was considered a preclusion criteria for further extensive (secondary) intervention based on the existing recommendations at the time of the study. Therefore, the primary objective of this study was to investigate the influence of MSI and DNA ploidy on recurrence in patients who were eligible for systematic surveillance after surgery. Thus, the results should be regarded for the population described (ie, patients with stage I-III CRC, aged <75 years, and curative surgery), and not for the general population of patients with CRC who undergo curative resection. Furthermore, we cannot exclude the possibility of a statistical Type II error based on a study sample that potentially was too small. Nonetheless, the truly population-based accrual of patients in this study was void of referral bias, such as that observed in trials investigating only patients who are eligible for adjuvant chemotherapy; therefore, the results should be applicable to other population-based patient cohorts aged <75 years.

The prognostic value of DNA ploidy and histogram descriptors has been a matter of debate, because some studies reported that aneuploidy was associated with a worse prognosis than diploid cancers, while other studies did not confirm those results. In line with this, high values of the other histogram descriptors that we analyzed also were associated with a worse outcome in certain (but not in all) tumors. None of the DNA ploidy features were prognostic in the current study. Some, if not all, of the reasons mentioned above for the lack of prognostic value for MSI also may hold for DNA ploidy features. Furthermore, although CIN to be a prognostic factor in patients with CRC, its precise role in tumor biology and its relation to MSI remains unclear.11 To the best of our knowledge, and as derived from a recent metanalysis,11 this study on MSI and DNA ploidy in patients with CRC is the second of only 2 publications that have examined these 2 features in combination, and the other study investigated only patients with colonic cancers who were recruited to adjuvant chemotherapy trials; those investigators demonstrated a prognostic significance for aneuploid cancers, but not for MSI.17 Obviously, the roles of MSI and CIN will require further investigation to clarify their precise role in CRC development, prognosis, and prediction.

In the current study, TNM classification was the most significant prognostic factor, and lymph node status was the single most predictive risk feature for any recurrence and for death from disease. Also, lymph node status was associated with the risk of developing distant metastasis, and patients who had a status of pN+ had an overall shorter time to the development of distant metastasis (all within 3 years) compared with patients who had a status of pN0. It is noteworthy that patients with MSI had a significantly increased risk of developing locoregional recurrence (25% of patients) with a nonsignificant trend toward a shorter time to locoregional recurrence. This is in line with results that demonstrated a reduced likelihood of metastasis present at the time of diagnosis in patients with MSI tumors28 as well as rare occurrences of MSI in hepatic CRC metastasis.37

The current findings, we believe, may have considerable clinical implications for planning systematic surveillance after curative surgery if they are validated by further studies. First, having a positive lymph node status indicates an increased risk of distant metastasis (ie, to the liver and lungs), and these patients preferably should undergo focused, imaging-directed surveillance, ie, with computed tomography scans of the abdomen and thorax at 1 to 3 years to detect curable metastasis. Improved stratification for imaging surveillance for patients who are at risk for distant metastasis could further improve the effect of image-based surveillance.38 Second, endoscopy-based surveillance may be more suitable for patients who have MSI tumors, because as they are less likely to develop distant metastases28 but more frequently develop intraluminal recurrence or metachronous CRC, as demonstrated in the current study. If implemented, this also could improve the lack of survival recently associated with endoscopic surveillance in CRC.25

In conclusion, a molecularly directed, tailored approach according to risk, time to recurrence, and site of recurrence could have great worldwide implications for a large number of patients with CRC who are undergoing surveillance after resection. Although MSI did not predict survival in the current patient population, it holds potential as a stratification tool for the risk of recurrent locoregional CRC after patients undergo curative surgery.

Conflict of Interest Disclosures

  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References

Supported in part by grant 165811/V50 from the Norwegian Research Council and by a grant from the Research Council of the Stavanger Health Trust.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures
  7. References