Introduction

Microsatellite instability (MSI) is now widely used as an indispensable biomarker in the field of clinical oncology, where it has been established that MSI straightforwardly reflects cellular defects in DNA mismatch repair (MMR). Defective MMR confers on tumour cells both insensitivity to fluoropyrimidines such as 5-fluorouracil (5-FU)1 and sensitivity to immune checkpoint inhibitors (ICIs)2,3,4. Therefore, MSI-positive tumours are expected to be resistant to the former and sensitive to the latter. 5-FU is one of the most frequently and widely used cytotoxic anticancer agents, and ICIs being regarded as a currently most promising target-based drug. Thus, the testing of MSI in tumour cells is now prerequisite in current cancer treatment strategies. For this purpose, five selected mononucleotide microsatellites are widely analysed using PCR and capillary electrophoresis5,6,7, although next generation sequencing (NGS) is also partly used. However, this PCR and capillary electrophoresis-based assay used worldwide is, in fact, not free of methodological problems8. In addition, molecular mechanisms of microsatellite destabilisation in eukaryotic cells are not limited to MMR but include polymerase proofreading9, 5’ flap processing10,11 and recombination12,13. It is thus expected that the microsatellite-unstable phenotypes in human cancer cells are not uniform but rather complex, and that the relationship of them to MMR deficiency is not straightforward. The established guidelines for MSI classification, however, simply utilise the frequency of instability in a set of microsatellites, i.e. MSI-H(high) and -L(low)14, and MSI-H has been established as a genuine phenotype reflecting defective MMR. These underlying complexities may cast a shadow on the accuracy and reliability of this biomarker in predicting the utility of fluoropyrimidines and ICIs.

Microsatellites are a class of short tandem repeat (alias simple sequence repeat (SSR)15) and comprised of short repeat units ranging from one to six-bp. Numerous microsatellites are distributed throughout the eukaryotic genome, and more than one million are mapped on the human genome. However, their physiological functions are unknown. Since DNA polymerases are highly prone to slip on the repetitive sequences such as microsatellites, strand misalignments leading to repeat length changes are frequently formed and repaired by MMR. However, microsatellites may be a hotspot for DNA replication errors including not only strand misalignments but also other modes of DNA strand damage. This may implicate the potential instability of eukaryotic microsatellites. Indeed, as mentioned above, several different mechanisms destabilise the repeat sequences, and not only MMR but also other repair systems stabilise them. We have previously shown the role of DNA polymerase proofreading in mononucleotide microsatellite stability9. More importantly, we have observed at least two qualitatively distinct modes of microsatellite alterations, i.e. Type A and Type B, in human tumours16. In case of dinucleotide microsatellites, Type A is defined as length changes within 6-bp (three repeat units), and Type B as more large-scale length expansions or contractions16 (see more detailed definitions in Materials and methods and Fig. 1). Type B alterations are remarkable and, therefore, readily detectable using common assay platforms, whereas Type A is more subtle and minor and, therefore, obviously prone to be overlooked by less sensitive assay techniques. Since Type B alterations are typically observed in MSI-H tumours, MSI-H/Type B has been directly connected to MMR deficiency. However, in a line of our observations using MMR-gene knock-out mice16 and genome-edited human cells17, we found that MSI directly caused by defective MMR is not Type B but Type A. In other words, MMR deficiency is sufficient for Type A, but not for Type B. The relationship between MSI-H/Type B and defective MMR is obviously not straightforward16,17. These findings strongly prompted us to explore a connection between Type A MSI and 5-FU insensitivity or ICI sensitivity in real cancer patients. In this study, we tested the former hypothesis. In order to sensitively detect Type A alterations in tumour DNA, we employed our precision and high-resolution assay techniques18,19. The results obtained strongly indicate that tumours exhibiting Type A MSI are significantly more resistant to 5-FU-based chemotherapy, although the cohort size was limited. Here, we report a thus far overlooked connection between 5-FU resistance and more subtle MSI.

Fig. 1
figure 1

The definitions of Type A and B MSI. The electrophoretic profiles of PCR amplified dinucleotide microsatellites are comprised of four peaks: + 2, 0, − 2 and − 4, and peak 0 corresponds to the correct size of template microsatellite sequence (Ref. 19) (a). When MSI occurs, there are several new minor microsatellite alleles with altered number of dinucleotide repeats, and the electrophoretic profiles are modified by them. Type A MSI is defined as an (i) appearance of new peaks within 6-bp (three repeats) (b, right and left) or a change in (ii) peak height ratio (PHR) above 0.15 between peak 0 and + 2 (b, left) or (iii) over 0.40 between peak 0 and − 2 (b, right). On the other hand, Type B is defined as an (i) appearance of new peaks over the length range of 6-bp (four or more repeat units) and/or an (ii) appearance of a new apex (c). Blue lines indicate parental (homozygous) alleles, and red lines altered MSI alleles. Observed electrophoretic profiles are shown by black interrupted lines.

Full size image

Materials and methods

Patients

Thirty colorectal cancer patients with surgically unresectable and objectively measurable metastatic lesions in the liver, who were consecutively diagnosed and treated in the Department of Gastrointestinal and Medical Oncology, NHO Kyushu Cancer Center from 2009 to 2015, were enrolled in this study. Patients received one or more cycles of fluoropyrimidine (5-FU or capecitabine) and oxaliplatin chemotherapy, i.e. mFOLFOX620,21 or CAPOX22, as a first-line treatment. These patients were ineligible for bevacizumab23, cetuximab24 or panitumumab25 because of the following medical conditions: bleeding, coagulopathy, history of cardiovascular diseases, ascites, unhealed wound and KRAS mutation etc. Ethical approval for this study was obtained from the institutional review board.

Microsatellite instability

Precision and high-resolution multi-fluorescence microsatellite analysis has been described in detail elsewhere18,19. Briefly, human dinucleotide microsatellite sequences, D2S123, D5S107, D10S194, D13S175 and D17S250, were amplified by PCR using TaKaRa Taq (Takara Bio Inc., Kusatsu, Japan) and primers labelled with a fluorescent dye, 6-FAM™ (6-carboxyfluorescein) or HEX™ (6-carboxy-2’, 4’, 7’, 4, 7, hexachloro-fluorescein) (Applied Biosystems, Thermo Fisher Scientific, Foster city, CA, USA). In addition to genomic DNA, artificially synthesised single-stranded oligodeoxynucleotides (ssODNs), the sequences of which are completely identical to the reference sequences except the repeat number, were also used as a template for PCR (see Sup. Figs. 1, 2). PCR products were electrophoresed in the ABI PRISM™ 3500 or 310 Genetic Analyzer (Applied Biosystems), and the data were processed using GeneScan™ and GeneMapper™ softwares (Applied Biosystems). Microsatellite repeat alterations are judged by the deviations from the basic electrophoretic profiles19. The most remarkable and readily scorable mode of microsatellite alterations is ( I ) Type B, which is defined as an ( i ) appearance of new peaks over the length range of 6-bp (insertions or deletions of four or more repeat units in dinucleotide microsatellites)16 and/or an ( ii ) appearance of a new apex (a highest peak in a peak cluster) (Fig. 1c). The Type B mode is universally observed in MSI-H tumours. On the other hand, the more subtle mode is ( II ) Type A, which was initially defined as an ( i ) appearance of new peaks within 6-bp (three repeats)16 (Fig. 1b). However, according to the results of the simulations in this study (see Sup. Figs. 1, 2), we have expanded the scoring criteria to include ( ii ) significant alterations in peak height ratio (PHR) between two peaks: above 0.15 between peak 0 and + 2 (Fig. 1b left) or over 0.40 between peak 0 and − 2 (Fig. 1b right). In this form of microsatellite assays, loss of heterozygosis (LOH) is also observable, because it causes a decrement of peak heights. However, some patterns of Type A alterations are theoretically indistinguishable from LOH, as discussed elsewhere26, and, therefore, not included as Type A in this study (see Sup. Table 1). Type A MSI positivity in tumours was defined as Type A alterations in more than one (≥ 40%) of five markers, according to the Bethesda guideline14.

Statistical analyses

Clinical characteristics were compared between patients with and without MSI by Fisher’s exact test or Bonferroni correction for categorical variables and Wilcoxon rank sum test for continuous variables. Tumour response to fluoropyrimidine and oxaliplatin chemotherapy was assessed according to the Response Evaluation Criteria in Solid Tumours (RECIST) version 1.127. One patient, patient code F06, was unable to be evaluated on response due to an unexpected early severe adverse event (intestinal perforation). Patient response, i.e. ‘complete response’ or ‘partial response’27, was compared between Type A MSI-positive and -negative patients by Fisher’s exact test. The diameters of five or less metastatic lesions were measured in each patient, and the maximal changes from the initial size were calculated, compared between patients with and without Type A MSI using the Wilcoxon rank-sum test and expressed as a waterfall plot (see Fig. 2). In this study, progression-free survival (PFS) was defined as the time from the start of first-line treatment to disease progression without or after response or death for any reason, overall survival (OS) as the time from the initial treatment to death. Survival time was recorded from the initiation of treatment until progression, last visit or death for PFS and last visit or death for OS. Data of patients were censored at the date of the last visit or the date on which the patient was last known to be alive unless disease progression and/or patient death occurred. As for four patients, F07, F13, F21 and F30, who underwent liver resection after fluoropyrimidine and oxaliplatin, data were censored at the day before the subsequent surgery and subjected to the PFS analyses. In one patient, F05, data were censored at the date of the last follow-up and subjected to PFS analysis. Survival curves were plotted according to the Kaplan–Meier method and compared using the log-rank test. In order to determine the significance of each clinicopathological variable as a prognostic factor, the Cox proportional hazards model was used. Cox regression analyses were performed using the eight variables: age, sex, ECOG PS, histopathology, tumour location, KRAS mutation, first-line chemotherapy and MSI, provided that the proportional hazards assumption holds. Multivariate Cox regression analyses were performed on the assumption that all the eight variables may be confounding factors, using a backward selection process, which excludes the variable with the largest p value in each round of analysis and is continued until only variables with p < 0.05 are obtained. Two-sided p values of < 0.05 were considered statistical significance. The above analyses were performed using JMP version 14.3.0 (SAS Institute, Inc., Cary, NC, USA) and/or R version 4.2.1 (The R Foundation for Statistical Computing, Vienna, Austria).

Fig. 2
figure 2

Tumour response to fluorouracil and oxaliplatin in colorectal cancer patients with Type A MSI+/− tumours. The diameters of metastatic liver lesions were measured in each patient who underwent first-line mFOLFOX6 or CAPOX, and the maximal changes from the initial size were compared and shown as a waterfall plot. Each bar represents the maximal relative change in the tumour size in each patient and is coloured according to the MSI status (Type A MSI+, red; Type A MSI-, blue) and chemotherapy regimen (mFOLFOX6, dark; CAPOX, light). The cut-offs for significant response and disease progression, indicated by dashed horizontal lines, are ≥ 30% decrease and ≥ 20% increase, respectively, which are based on the RECIST version 1.1. Data of Patient F06, who had an unexpected early severe adverse event, is not included in this figure.

Full size image

Results

More subtle microsatellite destabilisation in colorectal cancer

Before proceeding to the analysis of microsatellites on the genome in real patients, we re-verified the sensitivity and the quantitativity of our assay system. For this purpose, we artificially synthesised several single-stranded oligodeoxynucleotides (ssODNs) carrying the base sequences completely identical to those of human microsatellites with a variety in the repeat number and then used them as PCR templates19. Theoretically, MSI is in fact defined as an appearance of new alleles with different repeat numbers, and, therefore, most simply classified into two patterns: an appearance of alleles with longer or shorter repeats, which are finally expressed either as an appearance of new peaks or peak height alterations in capillary electrophoresis. We simulated the latter two patterns in vitro using the above ssODNs (Sup. Figs. 1, 2). The results clearly indicate that our system detects microsatellite length changes sensitively and quantitatively, and that not only small new peaks but also subtle changes in the relative peak height are indeed significant. According to these findings, we re-defined Type A MSI as an ( i ) appearance of new peaks within 6-bp (three repeats) or ( ii ) significant relative peak height changes (Fig. 1, Sup. Figs. 1, 2). Using this sensitive and quantitative system, we analysed tumour microsatellites in patients.

As a subject cohort, we selected colorectal cancer patients who had quantitatively assessable liver metastases and were treated with FOLFOX or CAPOX. The patient characteristics are summarised in Table 1. In these patients, tumour tissue specimens were collected and archived in the biobank in our hospital, and high molecular weight genomic DNA were extracted from them. Then, five dinucleotide microsatellites were analysed in tumour DNA, using the above assay technique. Microsatellites were indeed unstable in several tumours, and various alterations were observed. As expected, not only obvious new peak appearance (see examples in Sup. Fig. 3a) but also significant alterations in peak height ratio (PHR) (Sup. Fig. 3b) were observed. In addition to Type A alterations, Type B MSI (Sup. Fig. 3c) was also observed in one tumour (F19). The results are summarised in Table 2. In Type A MSI, there are some patterns of alterations that are not theoretically distinguishable from LOH (see Materials and methods). Therefore, these alterations are not scored as Type A in this study (see ‘A/LOH’ in Sup. Table 1). Finally, according to the Bethesda guideline14, Type A MSI positivity in each tumour was determined, i.e. tumours with Type A in two or more of five markers were judged as positive. Among 30 tumours, 11 (37%) were designated Type A MSI-positive. The tumour, F19, which exhibited both Type A and B alterations, was classified into the Type A-positive subgroup.

Table 1 Patient characteristics.
Full size table
Table 2 Microsatellite changes in colorectal cancer patients.
Full size table

Poor response to fluorouracil-based chemotherapy and poor survival in colorectal cancer patients with subtle microsatellite instability

In Type A MSI+ patients, seven (7/19, 37%) and four (4/11, 36%) were treated with mFOLFOX6 and CAPOX, respectively (Table 2). There were no significant differences in clinicopathological characteristics between patients positive and negative for Type A MSI (Sup. Table 2). According to the RECIST version 1.1, objective response after the first-line fluoropyrimidine and oxaliplatin treatment was analysed in Type A MSI-positive and -negative tumours, the rates of which being 20% (2/10 tumours) and 68% (13/19), respectively. Thus, the response to fluoropyrimidine and oxaliplatin was significantly poor in Type A MSI+ tumours (p = 0.021). Furthermore, we compared the maximal changes in the diameters of measurable metastatic lesions between Type A MSI-positive/negative tumours, and found that the Type A+ diseases were significantly stable in size after the first-line 5-FU-based chemotherapy (median maximal change -5.6% [interquartile range (IQR) -31.3 to +16.6] in Type A+ tumour vs. -43.4% [IQR -55.4 to -27.0] in Type A- tumour; p = 0.006) (Fig. 2).

As expected, median PFS in patients with Type A MSI- tumours was 7.1 months (95% confidence interval [CI] 5.4–9.2), whereas being 3.3 months (95% CI 0.7–4.6) in those with Type A+ tumours, the difference of which was statistically significant (p < 0.001) (Fig. 3a). Moreover, in addition to PFS, OS was significantly poor in Type A+ patients (median OS, 9.0 months [95% CI 1.3–24.4]; 25.0 months [95% CI 17.0–28.9] in Type A- patients; p = 0.009) (Fig. 3b).

Fig. 3
figure 3

Patient survival in Type A MSI+/− colorectal cancer after 5-FU-based first-line chemotherapy. In patients with Type A MSI-positive/-negative colorectal carcinomas who received first-line mFOLFOX6 or CAPOX, survival is expressed as a Kaplan–Meier curve and compared: progression-free survival (a) and overall survival (b). Cross symbols indicate censoring. Red line, Type A MSI+ patients; blue line, Type A MSI- patients.

Full size image

We also performed Cox proportional hazards regression analyses, in order to conclude whether or not Type A MSI is an independent predictor of patient survival (Table 3). Univariate analyses revealed that Type A MSI positivity is associated with OS (hazard ratio (HR) 2.79 [95% CI 1.25–6.22], p = 0.012) (Table 3). ECOG PS tended to be associated with OS, which was however not significant (HR 2.09 [95% CI 0.94–4.64]; p = 0.070). The other variables, such as age, primary tumour location and chemotherapy regimen etc., did not show significant associations either with PFS or with OS (Table 3). Furthermore, multivariate analyses revealed that Type A MSI is an independent predictor significantly associated with both PFS and OS in this pilot cohort (HR 10.51 [95% CI 3.20–34.50] for PFS, p < 0.001; HR 3.57 [95% CI 1.52–8.35] for OS, p = 0.003), although the CIs are broad reflecting the paucity of subjects and thus the statistical power of these analyses may therefore be limited (Table 3).

Table 3 Univariate and multivariate analyses using the Cox proportional hazard model for progression-free survival and overall survival.
Full size table

Discussion

Our precision and high-resolution assay technique detects microsatellite length changes sensitively and quantitatively. This technique has shown the previously unrecognised mode of MSI, Type A, which is more subtle than the widely acknowledged mode, Type B, that is typically observed in MSI-H (high) tumours. In this study, Type A MSI has been defined as an (i) appearance of new peaks within 6-bp (three repeats) or (ii) significant relative peak height changes (see Fig. 1, Sup. Figs. 1, 2). Since our previous observations using MMR-gene knock-out mice16 and genome-edited human cells17 suggest that Type A is a primary phenotype of MMR-defective cells, we explored a connection between Type A MSI and 5-FU insensitivity in thirty colorectal cancer patients treated solely with 5-FU-based regimens. The obtained results strongly indicate significant 5-FU resistance in tumours exhibiting Type A MSI. The quantitatively assessed response to the treatment, median PFS and OS were all significantly poorer in patients with Type A-positive tumours. In addition, Type A MSI was an independent predictor of patient prognosis. Although the size of the pilot cohort is extremely limited in this probing study, our findings may strongly suggest that more subtle Type A MSI better predicts 5-FU insensitivity in cancer patients.

Numerous studies of the relationship between the tumour MSI status and the efficacy of 5-FU-based chemotherapy have been done to date. However, the data reported in the literature are not conclusive, which is finely described and typically exemplified by a meta-analysis by Dogliotti E and colleague28. The lack of unity may derive from several problems. One obvious problem is the complexity in chemotherapeutic strategies. Although detailed mechanisms of 5-FU action have not been clarified to date, it has been well established in vitro that cells defective in MMR are resistant to 5-FU29,30,31. However, in current cancer chemotherapy, 5-FU is not used as a single agent but in combination with several other antineoplastic drugs. Furthermore, it is known that defective MMR modifies cellular sensitivities to some of these agents32,33. In advanced colorectal cancer, FOLFIRI34, FOLFOX35 and CAPOX22 regimens have been chiefly used, the former and the latter two of which include irinotecan and oxaliplatin, respectively. In the both agents, it has been suggested that defective MMR may influence the sensitivity of cancer cells to them32,33. Moreover, these 5-FU-based regimens are currently supplemented with several target-based drugs such as bevacizumab, cetuximab and panitumumab. These complex combinations of drugs have obviously made it difficult to observe effects of a single agent in real clinical settings. In this study, the subject cohort comprises colorectal cancer patients treated with mFOLFOX6 or CAPOX but without any other agents. This is the primary reason for the limited cohort size in this study. Needless to say, the subsidiary effects of oxaliplatin may have biased treatment response in the patients. The advantage of oxaliplatin treatment in MSI-positive tumours has indeed been reported36. This effect may, however, diminish or neutralise the observed 5-FU resistance in Type A MSI-positive tumours. On the other hand, effects of the target-based drugs are completely excluded in this cohort. We believe that the primary effects of 5-FU were, at least partially, observed in this study.

A more important problem is the connection between the MSI-H phenotype and defective MMR. The MSI phenomenon was first reported in Lynch syndrome (LS) patients37 carrying a germline mutation in MMR genes38,39. Since tumours arising in LS individuals exhibit the MSI-H phenotype, MSI-H has been straightforwardly connected to MMR deficiency. However, MSI-H is observed also in tumours occurring in non-LS individuals, which is not necessarily explicable by epigenetic silencing of or somatic mutations in MMR genes. Thus, the MSI-H phenotypes obviously appear heterogeneous. As mentioned above, we have observed Type A and Type B MSI in human cancer16. Microsatellite alterations of Type A and B modes are qualitatively distinct: the former is relatively subtle and minor, whereas the latter being more extensive and drastic. The mode difference may suggest their differential mechanistic origins. Indeed, it is known that microsatellite alterations occur not only by polymerase slippage and defective MMR but also by several different mechanisms, i.e. DNA polymerase proofreading9, 5’ flap processing10,11 and recombination12,13, as discussed above. One important finding in our previous observations is that defective MMR is sufficient for Type A but not for Type B alterations, although the latter mode is generally observed in MSI-H-positive tumours and has therefore been regarded as an established phenotype of cells defective in MMR. In human cells in which MMR gene mutations established as ones of LS kindred have been introduced using the CRISPR/Cas9 technique, the observed microsatellite changes were not Type B but uniformly Type A17, which suggests that Type A is a primary phenotype of MMR-defective cells. Type B may require previously unrecognised and additional mechanisms, and the relationship between MSI-H/Type B and defective MMR may not be straightforward as has been expected. Moreover, as a consequence of the above, MSI-H and Type A MSI do not overlap well (Fig. 4). We believe that these underlie the above mentioned confusion in the literature. The biological significance of Type A MSI has been confirmed not only by genetic engineering but also by clinical observations. Type A MSI has indeed been observed in various neoplastic diseases40,41,42,43,44,45,46,47 and shown to be associated with clinical features including therapeutic response and patient outcome44,46. In the present study, we have shown that more subtle Type A MSI may more directly and sensitively indicate MMR defects in tumour cells and therefore better predict their resistance to 5-FU in colorectal cancer patients.

Fig. 4
figure 4

A schematic visualisation of the relationship between MSI-H/L and Type A/B MSI. Our previously published data41 are expressed as a Venn diagram. Each area roughly corresponds to its frequency (Figures in the parenthesis are percentage.). Since both Type A and B alterations sometimes occur in a same set of microsatellite markers, their sets partially overlap, whereas MSI-H and -L are mutually exclusive, and the frequency of this Type A&B category is approximately 4%. Classical Lynch syndrome cases or those with epigenetically silenced MLH1 are included in this area (A). On the other hand, tumours with Type B MSI but without silenced MLH1 may not necessarily be MMR-defective (B). Type A MSI is generally observed in one or two of a set of five markers and, therefore, chiefly categorised as MSI-L (C) and sometimes as MSI-H (D). In a limited number of cases, Type B alterations are found in only one marker, i.e. MSI-L. However, this is not expressed in this diagram (E). Type A alterations are more subtle and minute than Type B and, therefore, prone to be overlooked without sensitive assay techniques or without being carefully observed (F).

Full size image

One marked drawback of this study is, needless to say, the extremely small cohort size. However, this was inevitable because patients treated solely with 5-FU-based regimens are limited, as discussed above. From this point of view, this study is a pilot to probe into a real patient population and to find out a previously unrecognised relationship between tumour MSI and patient response to 5-FU. Our findings may shed light on a thus far overlooked connection between more subtle microsatellite alterations and resistance to 5-FU-based chemotherapy and, therefore, warrant future studies of larger cohorts to confirm the found connection. Another problem to be addressed is the methodology to detect MSI. As shown in Sup. Fig. 3a and b, the microsatellite changes of Type A are indeed subtle and minute. They are not accompanied by an appearance of many new and high peaks or well separated clusters of them, which are typical in Type B alterations (Sup. Fig. 3c), but are comprised of an appearance of one small peak or of changes in the peak height ratio (Sup. Fig. 3a,b). In order to detect such subtle alterations, the assay system needs to be highly sensitive, quantitative and reproducible. For this purpose, our system utilises the precision and high-resolution assay technique18,19, which enables us to determine the absolute length of microsatellite PCR products with an accuracy of within one base. Furthermore, the conditions for PCR and capillary electrophoresis have been optimised in order to maximise the reproducibility. However, these implementations may not be possible in all assay sites. In this context, the DNA fragment analysis using capillary electrophoresis has a limit, although currently regarded as a standard methodology. On the other hand, next generation sequencing (NGS) has started to be used also for microsatellite analysis. The NGS approaches may have some advantages. The sequencing by synthesis (SBS) method is free from polymerase slippage, and, therefore, subtle and minute microsatellite alterations may be precisely detected, although the assay process partly includes PCR. However, the algorithms to call sequence variations have been developed only to pick up large-scale microsatellite length changes such as those in MSI-H/Type B. More subtle and minor Type A alterations have been similarly overlooked as in the common capillary system. New algorithms for SBS-based microsatellite analysis need to be exploited. Nevertheless, considering the costs in NGS, the capillary electrophoresis assays may still have advantages, if being done more precisely, because they are easily accessible, fast and more economical. Further efforts are required to establish MSI as a more reliable biomarker and, consequently, to truly personalise 5-FU-based chemotherapy for more effective treatment of cancer patients.

Conclusions

More sensitive MSI assays reveal the Type A mode, which is more subtle than the widely acknowledged mode typically observed in MSI-H (high) tumours, i.e. the Type B mode. We previously observed Type A MSI in MMR-gene knock-out mice16 and genome-edited human cells17, which suggests that Type A is a primary phenotype of defective MMR. In this study, we explored a connection between Type A MSI and 5-FU insensitivity in colorectal cancer patients treated with 5-FU-based regimens. The obtained results strongly indicate significant 5-FU resistance in Type A-positive tumours. Although the cohort size of this probing study is extremely limited, our findings may strongly suggest that Type A MSI better predicts 5-FU insensitivity in cancer patients. Our findings warrant future studies of larger cohorts.