Precision medicine for adjuvant chemotherapy of resected colorectal cancer

Abstract Colorectal cancer (CRC) is the most common cancer and the second leading cause of cancer death in Japan. Surgical resection is the only curative option for localized disease. However, undetectable micrometastases remaining after curative surgery may cause disease recurrence. Adjuvant chemotherapy aims to eradicate these micrometastases to improve the cure rate. Unfortunately, few reliable prognostic and predictive markers are available that identify patients at high risk for CRC during early‐stage disease. However, promising biomarkers may become available in the near future. Such biomarkers provide information for stratifying a patient's risk and for selecting the optimal treatment. Here, we provide an overview of current relevant prognostic and predictive biomarkers applicable to adjuvant treatment of early‐stage CRC and focus on the future of this field.


| INTRODUC TI ON
Colorectal cancer (CRC) is the most common cancer and the second leading cause of cancer death in Japan, with more than 130 000 new cases diagnosed each year. 1 Surgical resection is the only curative option for localized disease, and its outcome is most closely associated with the extent of disease upon presentation. However, disease recurrence is caused by undetectable micrometastases that persist after curative surgery. Adjuvant chemotherapy aims to eradicate these micrometastases to improve the cure rate.
Disease stage is the most important pathological prognostic factor after surgery. For example, the Japanese Multicenter Registry reported recurrence rates after potentially curative surgery of approximately 15% and 32% for stages II and III CRC, respectively. 2 Furthermore, a phase III study of patients treated in Japan found that the 5-year disease-free survival (DFS) rates of patients with stages IIIA, IIIB, and IIIC were 90.4%, 74.1%, and 58.9%, respectively. 3 As of early 1990, 5-fluorouracil (5-FU)-based regimens were generally accepted as a standard adjuvant regimen for patients with stage III and selected patients with stage II colon cancer with 5%-10% improvement in absolute survival. [4][5][6][7] Furthermore, the addition of oxaliplatin lengthens DFS and OS of patients with stage III CRC. [8][9][10] The benefit of these adjuvant chemotherapies is most clearly demonstrated by the approximately 30% relative risk of recurrence of stage III disease. 11,12 However, the importance of adjuvant chemotherapy in the treatment of stage II disease is unclear.
Several clinical trials found that adjuvant chemotherapy of resected stage II colon cancer confers a minimal benefit upon OS. 3,13,14  Are these patients candidates for additional therapy? 2,15 In this era of precision medicine, molecular characterization of tumors is essential for selecting a therapeutic strategy. Consequently, the identification and standardization of prognostic and predictive molecular biomarkers for cancer are becoming increasingly relevant.
Here, we summarize knowledge of candidate prognostic and predictive biomarkers for adjuvant treatment of stages II and III CRC (Table 1), with a focus on the future of this critically important field.

| Tumor budding
Another finding may contribute new insights to postoperative adjuvant therapy for stage II CRC. The clinical value of the tumor budding status as a tumor-associated prognostic factor was addressed by the prospective, randomized controlled SACURA trial 3 This trial evaluated the superiority of adjuvant treatment with oral tegafururacil compared with surgery alone for stage II colon cancer. 18 Tumor budding is defined as a single tumor cell or a cell cluster consisting of ≤ 4 tumor cells. 19 The International Tumor Budding Consensus Conference was held to reach an agreement on an international, evidence-based standardized scoring system for tumor budding in CRC. 19 High tumor budding status is significantly associated with a lower 5-year recurrence-free survival (RFS) rate, and multivariate analyses of RFS revealed that budding status is an independent prognostic factor. 18

| G ENE TI C ALTER ATI ON S
Recent studies summarized below focus on identifying specific genomic mutations that serve as predictive biomarkers of adjuvant chemotherapy after surgical treatment that can be employed to individualize treatment. Several key mutations in CRC are important for its initiation, progression, metastasis, and response to therapeutics.

| Microsatellite instability (MSI)
Microsatellite instability is caused by a deficiency in DNA mismatch repair (MMR), resulting in the accumulation of mutations in DNA.

DNA-MMR deficiency is detected in approximately 15% of sporadic
CRCs. 20 A germline mutation that inactivates MMR genes may lead to Lynch syndrome, which is a common hereditary disorder that predisposes patients to CRC. Young age and a positive family history of CRC are risk factors for Lynch syndrome. MSI/dMMR tumors are characteristically located on the right side of the colon, exhibit mucinous histology with tumor-infiltrating lymphocytes, and are most frequently diagnosed as stage II. 21 Furthermore, most studies show that MSI/dMMR is an independent favorable prognostic factor of survival of patients with CRC. [22][23][24] Ribic et al 23 published the first report of differential benefit conferred by fluorouracil-based adjuvant chemotherapy upon patients with stages II and III colon cancers with microsatellite-stable and MSI-low tumors compared to those with MSI-high tumors. These findings are consistent with those of a systematic review of seven studies that stratified survival of patients with CRC according to MSI status. 25 In contrast, two analyses confirm the prognostic significance of dMMR but not its predictive capability. 26

| KRAS
KRAS is a proto-oncogene encoding a 21-kDa GTP-binding protein that regulates cellular responses to numerous extracellular stimuli.

| TP53
TP53 encodes a tumor suppressor that initiates cell cycle arrest, apoptosis, DNA repair, and the inhibition of angiogenesis. 44  alterations. [45][46][47] A systematic review found that abnormal TP53 is associated with an increased relative risk of death, regardless of whether immunohistochemistry or DNA mutation analysis was used. 46 In clinical studies of adjuvant chemotherapy-treated and untreated groups, patients with stage III CRC whose tumors overexpressed TP53 experienced significantly shorter survival following 5-FU-based chemotherapy than patients whose tumors did not express TP53 with detectable alterations. 48,49 However, other studies of patients with colon cancer failed to demonstrate correlations between TP53 alterations and the benefit of adjuvant therapy. 50,51

| IMMUNOSCORE
Accumulating evidence suggests that the tumor microenvironment is required for disease progression and tumor resistance to chemotherapy, 56,57 such that the assessment of tumor-infiltrating lymphocytes for prognostication and prediction of benefit from adjuvant chemotherapy is critically important. 58 The Immunoscore is a scoring system based on the densities of CD3 + and CD8 + T cells at the tumor center and the invasive margin, which is determined using immunohistochemistry and quantified using digital pathology.
The Immunoscore was defined in a large international validation study of 2681 patients with stages I-III colon cancer. 59 Patients with a high Immunoscore are at lowest risk of recurrence (HR, high vs low Immunoscore = 0.20, 95% CI = 0.10-0.38, P < .0001). These findings were independently confirmed using internal and external validation sets. Among patients with stage II colon cancer, the Immunoscore accurately assessed relapse risk regardless of MSI status and identified patients for whom only surgery would be a sufficient treatment option. Furthermore, the ability of the Immunoscore to predict OS was superior to that of existing tumor-risk parameters. Interestingly, the immune infiltrate varied widely between patients; 21% of patients with MSS tumors had high Immunoscores. These data suggest that factors such as characteristics of tumors, the microenvironment, and genetic and epigenetic alterations may affect the quality and density of the immune infiltrate.
Recently, the IDEA France cohort study evaluated the associations between the Immunoscore and DFS after 3 or 6 months of oxaliplatin-based adjuvant chemotherapy administered to patients with stage III colon cancer. 60 Furthermore, patients with an intermediate or high Immunoscore in the clinical low-(T1-T3, N1) and high-risk (T4 and/or N2) groups derived a significant benefit from the 6-month mFOLFOX6 regimen compared with the 3-month treatment.

| Gene expression signature
Gene expression is intimately linked to cellular phenotype and tumor behavior and is extensively used to identify biologically homogeneous subtypes of a disease. However, with a few exceptions, many single-gene expression markers make it difficult to predict the risk of recurrence in patients with CRC. Therefore, studies were conducted to improve predictive accuracies using combinations of multiple biomarkers. Prognosis predictions using gene expression signatures (OncotypeDX, Coloprint, GeneFX, OncoDefender-CRC, and ColonPRS) are currently available for stages II and III CRC.
OncotypeDX is a 12-gene RT-PCR assay to identify genes that predict recurrence and treatment effects of stages II-III colon cancer. After its clinical significance was initially introduced in an analysis of breast cancer, 66 the assessment was applied to patients with CRC. For example, one study assessed the association between colon cancer recurrence and expression of multiple genes detected using paraffin-embedded tumor tissues acquired from patients with stage II or III colon cancer treated in four independent trials of adjuvant therapy that included surgery-alone and 5-FU-adjuvant chemotherapy arms. 67 Among an initial 761 candidate genes, 12 were selected according to their independent association with recurrence in each of the studies. The analysis, which includes five reference genes, found that seven and six genes are significantly associated with recurrence and treatment benefit, respectively.
The prognostic accuracy of Oncotype DX was validated in an analysis of data from the prospective QUASAR trial 68 of stage II colon cancer. Among the 711 patients enrolled in the surgery-alone arm of the study, the seven-gene recurrence score was significantly associated with recurrence risk 3 years after surgery (P = .004).
Additional validation involved a separate analysis of data of the CALGB 9581 trial 69 of stage II colon cancer and the SUNRISE study 70 of stages II-III colon cancer.
The predictive value of this treatment score could not be validated using the QUASAR sample. An additional prospective study was designed for clinical validation of patients in the NSASBP C-07 trial with stages II-III colon cancer who received 5FU alone or oxaliplatin-based therapy. 71 This study found that Oncotype DX is not a predictor of oxaliplatin treatment efficacy and therefore cannot identify patients for whom oxaliplatin is not beneficial. However, this study provides evidence that patients with higher recurrence scores may derive an absolute benefit from oxaliplatin vs those with a low recurrence score.

| CDX2 expression
CDX2 is a highly sensitive and specific marker of adenocarcinomas of intestinal origin. 75 CDX2 is a tumor suppressor, and its expression is frequently downregulated in CRC. 76  there is a significant benefit associated with adjuvant chemotherapy among CDX2-negative patients than among CDX2-positive patients.
These results suggest that CDX2 expression serves a prognostic biomarker for stages II and III colon cancer that may be effectively treated using adjuvant chemotherapy. Although encouraging, these results are derived from retrospective analyses of patient cohorts and pooled data sets, and therefore require validation by prospective randomized trials before CDX2 expression can be incorporated as a biomarker into routine clinical practice.

| FUTURE PER S PEC TIVE S
The remarkable progress in developing artificial intelligence (AI) technologies such as machine learning and deep learning enables multimodal analyses of big omics data. AI can therefore be applied to developing a platform to conduct multimodal analyses of biomarker complexity to potentially accelerate the realization of the promise of precision medicine. For example, machine learning approaches using radiomic features predict the response to chemotherapy in neoadjuvant settings. [79][80][81] Furthermore, other investigators analyzed machine-learning techniques combining with molecular-based parameters, which established clinicopathological features to identify the prognostic values of biomarkers. 82,83 The next promising field is the microbiota. The gut microbiota affects immunity, metabolism, and tissue development. [84][85][86] Recently acquired evidence suggests that specific microbiomes such as those comprising Fusobacterium are associated with the response to chemotherapy of patients with gastrointestinal cancers. 87,88 In vitro studies show that the combination of Fusobacterium and tumor cells influences the trend of chemoresistance to 5-FU and oxaliplatin. 87 Accordingly, the current phase III POLEM trial is designed to investigate the use of avelumab plus fluoropyrimidine-based chemotherapy as adjuvant treatment for stage III colon cancers with dMMR or POLE mutations. 93 Hopefully, these approaches will further personalize the treatment options in the adjuvant setting of CRC.
The regular use of aspirin after a diagnosis of colon cancer is associated with a superior clinical outcome. Interestingly, this effect of postdiagnosis aspirin use on survival appears to differ according to the PIK3CA mutational status. 94,95 A recent study shows that the association of aspirin use with CRC survival is significant for patients with CD274-low tumors vs CD274-high tumors. 96 The ongoing Add-Aspirin phase III, multicenter, double-blind, placebo-controlled randomized trial aims to determine the effects of aspirin on patients with CRC who have undergone potentially curative treatment. 97 Its available feasibility results indicate that aspirin is well tolerated after radical cancer therapy. 98 Another phase III trial is ongoing in Japan (JCOG1503C). 99

| CON CLUS IONS
During the last decade, numerous efforts have been made to develop precision medicine for patients with early colon cancer; however, biomarker-guided adjuvant treatment options are limited. Furthermore, novel drugs with specific targeting activity are not effective for treating patients with early colon cancer. The only standardized and efficacious treatment is 5-FU plus oxaliplatinbased chemotherapy. Effective molecular biomarkers are therefore required to identify optimal treatment strategies for managing patients who will benefit from adjuvant chemotherapy. Among these biomarkers, MMR status, Immunoscore, CDX2, and ctDNA, as well as others, will help predict a specific prognosis and response to adjuvant chemotherapy of patients with CRC. Although there are numerous concerns over the accuracies of the studies of most cancer treatments, we are rapidly advancing along the path leading to improving a cancer patient's prognosis.

ACK N OWLED G EM ENTS
We thank Edanz Group (www.edanz editi ng.com/ac) for editing a draft of this manuscript.

D I SCLOS U R E
The authors declare no conflict of interest.