Colorectal cancer under 20 years old: a retrospective analysis from three tertiary hospitals

Purpose Colorectal cancer (CRC) rarely occurs in children and adolescents. This study aimed to perform a retrospective analysis and disclose more detailed information about CRC in patients under 20 years old. Methods Medical records of CRCs in patients under 20 years old referred to three tertiary hospitals in China from September 2000 to July 2019 were retrospectively reviewed. Clinicopathological characteristics, treatment processes and laboratory findings were summarized and treatment outcomes and prognostic factors were analyzed. Results A total of 33,394 CRC medical records were analyzed, and we identified seventy (0.21%) CRCs in patients under 20. The most common primary tumor location was the left hemicolon (35.7%). The prominent pathological types were mucinous adenocarcinoma (22.9%) and signet ring cell carcinoma (22.9%). Nearly half (47.1%) of the patients presented with distant metastasis at diagnosis. The fractions of patients with deficient mismatch repair (dMMR) protein expression and microsatellite instability-high (MSI-H) were 23.8% (5/21) and 71.4% (5/7), respectively. Forty-four patients underwent radical surgery. Fifty-five patients received chemotherapy and six patients received radiotherapy. One dMMR/MSI-H rectal cancer patient received immunotherapy and achieved a clinically complete response. The median overall survival (OS) time was 80 months. The 3-year and 5-year OS rates were 61.8% and 57.2%, respectively. An absence of distant metastasis was a favorable factor for OS. For stage II/III CRCs, classic adenocarcinoma and radical surgery were favorable factors for OS. For stage IV CRCs, primary location at the colon was a favorable factor for OS. Conclusion Child and adolescent CRC patients are likely to have distant metastasis, undifferentiated, left hemicolon location, and a dMMR/MSI-H phenotype at diagnosis. Additional efforts are needed to improve their survival outcomes.


Introduction
Colorectal cancer (CRC) is the third most common cancer in incidence and ranks second in terms of cancer-related mortality (Bray et al. 2018). Adolescent and young adult (AYA) cancers accounted for 5% of all newly diagnosed invasive cancers in the United States between 2011 and 2015 (Close et al. 2019). We have witnessed an overall trend of decreased incidence of CRC over the past few decades as a result of using colonoscopy and other screening modalities in the older population (Connell et al. 2017). However, CRC incidence in young individuals is steadily rising (Kasi et al. 2019). AYA CRCs present with more advanced stage, poorer cell differentiation, and higher prevalence of signet ring cell histology, and the primary tumors are commonly located in the left side of the colon at diagnosis. Although CRC accounted for only 2.5-3.5% of cancer incidence in adolescents from 15 to 19 years old from 2011 to 2015, the 5-year survival rate was lower than in many other cancers (Close et al. 2019). Both the incidence and mortality rates of CRC are increasing in China in recent decades (Arnold et al. 2017). There are few reports on child and adolescent CRCs in patients under 20 (Kaplan et al. 2019;Indini et al. 2017;Sultan et al. 2010;Yang et al. 2010), especially in China (Du et al. 2015). To better characterize the clinical features, treatment strategies, and outcomes of CRC in children and adolescents, we performed an analysis of CRCs from three tertiary hospitals in South China.

Patient population
Patients with pathologically diagnosed CRC in three tertiary hospitals (Sun Yat-sen University Cancer Center, the First Affiliated Hospital, and the Sixth Affiliated Hospital of Sun Yat-sen University, Guangzhou, China) from September 2000 to July 2019 were initially considered. Eligibility criteria included the following characteristics: (1) pathologically diagnosed classic adenocarcinoma, mucinous adenocarcinoma (MA), or signet ring cell carcinoma (SRCC); and (2) under 20 years of age. Patients were excluded using following criteria: (1) prior history of other malignancies; and/or (2) severe hematopoietic, heart, lung, liver, or kidney dysfunction. Clinical characteristics, treatment process, laboratory findings, and survival outcomes were captured from the medical records and follow-up systems.

Ethics
The present study was performed in accordance with the ethical standards as presented in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards and was approved by the Ethics Committee of Sun Yat-sen University Cancer Center with a waiver of informed consent because this research was retrospective and did not involve accessing any patient identification data.

Tumor location
The colon comprises the left hemicolon, the right hemicolon, and the transverse colon. The left hemicolon includes the left flexura, the descending colon, and the sigmoid, and the right hemicolon comprises the cecum, the ascending colon, and the right flexura. The rectum compromises the anus to 15 cm above the anocutaneous line.

Staging and tumor biomarkers
All tumors were staged or restaged according to the 8th edition of the American Joint Committee on Cancer (AJCC) TNM staging system. Carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) were assessed at diagnosis or when patients were referred to one of the above three hospitals.

Tumor molecular characterization
Primary tumors were used for tumor molecular analysis. Mismatch repair (MMR) proteins, including MLH1, MSH2, MSH6, and PMS2, were assessed by immunohistochemistry (IHC). Tumors were classified as MMRdeficient (dMMR) if loss of one or more of the proteins was shown. Microsatellite status was detected by nextgeneration sequencing (NGS) or polymerase chain reaction (PCR) of five microsatellite markers, including BAT25, BAT26, D5S346, D2S123, and D17S250. MSI-high (MSI-H) was defined as presence of ≥ 30% mutations as detected by NGS or ≥ 2 microsatellite marker instability by PCR. K-ras mutation was assessed by NGS and/or PCR.

Follow-up
All patients were followed at 3-month intervals during the first 2 years, at least every 6 months thereafter for an additional period of 3 years, and then once a year until March 2020. Overall survival (OS) was defined from the date of diagnosis until death from any cause or was censored at last follow-up.

Statistical analysis
Continuous data are presented as the median with range, and categorical data are presented as proportions (%). Proportions were compared using a χ 2 test. Survival rates were compared with the log-rank test. A two-sided p value of < 0.05 was considered statistically significant. Covariates with p value < 0.05 by univariate analysis were subjected to multivariate analysis. All statistical analyses were performed using IBM SPSS Statistics 25.0 (IBM Co, Armork, NY, USA) and GraphPad Prism version 8.3.0 (GraphPad Software, Inc., La Jolla, CA, USA).

Patient characteristics
A total of 33,394 CRC medical records were analyzed, and we identified 70 CRCs (0.21%) in patients who were under 20 years old with histological diagnoses of adenocarcinoma, some of whom were referred from other hospitals. The characteristics of the patients are shown in Table 1. Of the 70 patients, 28 were diagnosed from September 2000 to December 2009, and the other 42 patients were diagnosed between January 2010 and July 2019. The median age at diagnosis was 18 (range, 8-20), with 30 patients (42.9%) being less than 18 years old. The proportion of male and female patients was 60.0% and 40.0%, respectively. Eleven patients (15.7%) had a family history of malignant tumor, and only 3 (4.3%) had a family history of CRC (one was a parent).

Pathology and staging
Classic adenocarcinoma, MA and SRCC were diagnosed in 38 patients (54.3%), 16 patients (22.9%), and 16 patients (22.9%), respectively. Almost all MAs and SRCCs were in the colon except four rectal MAs and one rectal SRCC. Thirty-three patients (47.1%) presented with distant metastasis in the abdominal and pelvic cavity (n = 16), liver parenchyma (n = 9), peritoneal surface (n = 8), lung (n = 3), distant lymph nodes (n = 3), bone (n = 2), and ovarian parenchyma (n = 1). Regional lymph node status was available in 51 patients, 41 of whom had positive regional lymph nodes. None of the patient presented with stage I disease. The number of patients with stage II and III disease was 9 and 25, respectively. Staging in 3 patients was unclear, but 2 of them were stage II or III. There were 36 patients (51.4%) with locally advanced CRC (LACRC) and 33 patients (47.1%) with distant metastasis (see Table 1).
MMR proteins were examined in 21 patients. Five patients (23.8%) exhibited dMMR. MLH1 and PMS2 were both negative in two rectal cancer patients, MSH2 and MSH6 were both negative in another rectal cancer patient and a left hemicolon cancer patient, and PMS2 and MSH2 were both negative in one patient with transverse colon cancer.
Microsatellite status was assessed in seven patients. Five patients exhibited MSI-H, three of whom had rectal cancer and two of whom had left hemicolon cancer. Microsatellite stability (MSS) was found in one right hemicolon cancer patient and one transverse colon cancer patient.
Five patients were detected with both MMR and MSI. Two were dMMR and MSI-H; one was pMMR and MSS. MMR and MSI status was inconsistent in two patients. One was pMMR but MSI-H, and the other was dMMR but MSS. Detailed MMR and MSI information in these five patients is shown in Table 2.

Treatment and curative effect
Twenty-five locally advanced colon cancer patients underwent radical surgery. Three received neoadjuvant treatment,  with a FOLFOX (5-fluorouracil, oxaliplatin plus leucovorin) regimen in 2 patients and neoadjuvant radiotherapy in one patient. Downstaging was not observed in these three patients, but all of them received adjuvant chemotherapy with FOLFOX regimen after surgery. Twenty patients did not receive neoadjuvant treatment, only adjuvant chemotherapy, with 12 patients receiving CAPOX (capecitabine plus oxaliplatin) and 8 patients receiving FOLFOX. The remaining two patients underwent surgery only.
In 11 locally advanced rectal cancer patients, only 4 (36.4%) received neoadjuvant therapy. One received CAPOX with Avastin combined with radiotherapy as part of a clinical trial, and he experienced pathological complete remission (pCR) after surgery. One received CAPOX combined with radiotherapy, and another received FOLFOX chemotherapy only. The pathology of these latter two patients after surgery suggested T and N downstaging. The last patient received CAPOX combined with radiotherapy. However, her tumor did not exhibit obvious regression and could not be completely removed, so palliative colostomy was performed. dMMR and MSI-H were confirmed after one cycle of capecitabine oral administration for a distal T3N1M0 rectal cancer patient, demonstrating that immunotherapy alone was applied in her case. Pembrolizumab was given for six cycles with the addition of Ipilimumab in the second and third cycles. Clinical complete response (cCR) of the tumor was found after the sixth cycle of immunotherapy. A watch and wait strategy was suggested for her, and Pembrolizumab was prescribed continuously for another four cycles.
For the 33 patients who presented with distant metastasis, 5 patients did not receive any anti-cancer treatment, and another 3 patients had palliative chemotherapy only. The remaining 25 patients underwent surgery, with 10 having radical resection of primary and metastatic foci, and 15 underwent palliative surgery. Among these 25 patients, 4 received preoperative chemotherapy, 15 received postoperative chemotherapy, 1 received adjuvant radiotherapy, and 1 received chemotherapy both before and after surgery.
In summary, 44 patients underwent radical surgery and 16 patients underwent palliative surgery. Fifty-six patients received chemotherapy, and 6 patients received radiotherapy.

Survival analysis
The median follow-up time of patients was 31 (0-203) months. Among all 70 patients, 4 (3.3%) were lost to follow-up. The median OS was 80.0 months for all patients. The 3-year and 5-year OS rates were 61.8% and 57.2%, respectively. When we investigated the association between clinical factors and outcomes using univariate analysis, only stage II/III (p = 0.001) was significantly associated with favorable survival (Table 3, Fig. 1).
We then explored factors affecting the survival of locally advanced and distant metastatic CRCs.
The median OS did not reach for locally advanced CRCs. The 3-year and 5-year OS rates were 77.8% and 73.5%, respectively. Classic adenocarcinoma (p = 0.019) ( Table 4, Fig. 2) and radical surgery (p = 0.025) (Table 4) were significantly associated with favorable OS in univariate analysis. The 3-year and 5-year OS rates of classic adenocarcinoma were 86.4% and 86.4%, respectively, and the 3-year and 5-year OS rates of radical surgery were 80% and 75.6%, respectively. When the aforementioned factors were compared by a multivariate model using Cox proportional hazards regression analysis, radical surgery had the lower hazard ratio (HR) of 0.042 (95% CI 0.003-0.584; p = 0.018), and classic adenocarcinoma had an HR of 0.182 (95% CI 0.045-0.734; p = 0.017) ( Table 4).
With respect to distant metastatic CRCs, the median OS was 23.4 months, and the 3-year and 5-year OS rates were 44.7% and 39.7%, respectively. Only primary tumor location in the colon was significantly associated with favorable survival (p = 0.044) ( Table 5, Fig. 3a). The 3-year and 5-year OS rates of colon cancer were 56.7% and 49.6%, respectively. The median OS in patients who underwent radical surgery was much longer than that in patients who did not (92.83 months vs. 9.5 months), although the p value was not significant (p = 0.053) ( Table 5, Fig. 3b).  there were only 19 CRCs in patients who were between 10 and 20 years old (Du et al. 2015). No large cohort of child and adolescent CRCs has been previously reported from China. We only identified 70 patients under 20 from September 2000 to July 2019 in three hospitals. The present study is one of the largest cohorts to examine the clinicopathological and prognostic features of this disease. The latest CRC statistics show that distant-stage diagnoses in patients younger than 50 years, 50-64 years, and 65 years and older are approximately 26%, 23%, and 19%, respectively (Siegel et al. 2020). However, nearly half of patients were stage IV in the present cohort. Presenting signs and symptoms in CRC are often vague and nonspecific, and no routine screening has been suggested. Child and adolescent CRC may have severe delays in diagnosis and/ or possess intrinsically more aggressive behavior. Therefore, additional effort is needed for early recognition of this malignancy.
In previous reports, histological types of MA and SRCC are approximately 10% and 1% among all CRCs (Verhulst et al. 2012;Gopalan et al. 2011;Nitsche et al. 2013;  Hyngstrom et al. 2012), but the proportions of MA and SRCC are higher in children and adolescents than in adults (Kaplan et al. 2019). In the current study, MA (22.9%) and SRCC (22.9%) histology was observed much more frequently than reported in adults. Several studies have indicated poor prognosis in patients with mucinous histology (Verhulst et al. 2012;Hyngstrom et al. 2012;Sung et al. 2008). However, other studies found that neither MA nor SRCC was an independent predictor of decreased survival Nitsche et al. 2016). Herein, we found that MA and SRCC indicated poor prognosis for LACRC. Due to the rarity of CRC in patients from such young age groups, clinical management and treatment approaches are generally decided according to experiences from the management of adult patients. The NCCN and ESMO guidelines both recommend multimodal treatment of locally advanced rectal cancer (LARC), involving neoadjuvant concurrent fluoropyrimidine-based chemotherapy with pelvic radiation, total mesorectal excision (TME), and adjuvant chemotherapy. In our study, among the nine LARC patients who underwent radical surgery, only three received neoadjuvant chemoradiation. Infertility might be the primary reason for omitting radiotherapy in LARC patients. Radical surgery was not successfully performed in two patients, both of whom died within 2 years after diagnosis.
For metastatic patients, surgery and chemotherapy were the most mainstream treatments in the present study and five patients did not receive any anti-cancer treatment for some reason. Radical surgery, which occurred in approximately 33.3% of patients in our cohort, may represent the only method to prolong survival for them. A previously study reported 5-year survival rates in patients diagnosed with distant metastasis from the ages of 20-49, 50-64, and 65 and older of 21%,16%, and 10%, respectively (Siegel et al. 2020). In our study, the 5-year OS rate of CRCs in patients under 20 was 39.7%, which is much higher than in patients older than 20. The higher rate of radical surgery may contribute to this increased rate of survival.
Health providers should do further investigations, such as genetic studies, to better understand the disease and to identify age-appropriate solutions. Clinicians and patients should be more engaged because more drugs are currently available, such as molecular targeted drugs and a multidisciplinary therapy model, which is implemented in most hospitals. dMMR was represented in 15-20% of stage II/III CRCs and in approximately 5% in the metastatic setting (Auclin et al. 2017). Five (23.8%) patients were dMMR, and 4 (66.7%) patients were MSI-H in our cohort, which is far higher than rates observed in adults. As dMMR/MSI-H is a major risk factor for CRC and involves a high percentage of dMMR/MSI-H in children and adolescent CRC patients, it is necessary to assess dMMR/MSI-H status when children and adolescents are diagnosed with CRC. However, MSI/MMR status was only available in a small number of patients, and more patients are needed for further validation. There was also a discrepancy between the results of immunohistochemistry and molecular detection in one patient who was dMMR but MSS and in one patient who was pMMR but MSI-H in our cohort. There are some reasons for this discrepancy. Some MMR gene variants do not affect the structure of their corresponding antigens and allow for retained normal MMR protein expression, but MMR proteins cannot recognize and repair DNA damage. Deficiencies in certain MMR proteins, such as MSH6, are not sufficient to result in MSI. Assessment of MSI and dMMR status discrepancy can be associated with a falsepositive or a false-negative. Immune checkpoint blockades (ICBs), such as anti-PD-1 antibody and anti-CTLA-4 antibody, are effective in MSI-H/dMMR metastatic CRC. Recent studies in the refractory CRC setting have led to US Food and Drug Administration approvals for Pembrolizumab and Nivolumab (with or without Ipilimumab) for MSI-H/dMMR metastatic CRC.
In this study, the patient who received ICBs as neoadjuvant treatment achieved cCR and did not have surgery. Recently, PD-1 plus CTLA-4 blockade was demonstrated as highly effective in early-stage dMMR colon cancers (Chalabi et al. 2020). For the higher percentage of dMMR/ MSI-H in child and adolescent CRCs, these patients may greatly benefit from immunotherapy. However, only a few patients have been reported (Chalabi et al. 2020;Zhang et al. 2019;Liu et al. 2020), and more data are needed for confirming and determining the best combination of immunotherapy with or without chemotherapy and/or radiotherapy to achieve a higher complete response rate and longer OS. There are some ongoing clinical trials attempting to add ICBs as neoadjuvant treatments for LACRC. In addition, we initiated a phase II clinical trial to combine PD-1 blockade with chemoradiation for dMMR/MSI-H unresectable CRCs. Genetic testing should be performed upon initial diagnosis for all child and adolescent CRCs, and immunotherapy is a promising modality for facilitating survival benefits in these patients.
There are some limitations to our study. This study is not a prospective cohort and has some of the inherent inadequacies of retrospective investigations. For example, molecular detections were available in only a small number of patients. In addition, the cohort covered nearly 20 years, thus a proportion of patients were lost, which may cause bias during analysis. Consequently, our observations warrant further consideration and validation in a larger patient series.

Conclusions
Child and adolescent CRCs are prone to having distant metastasis, poor pathological type, left hemicolon location, and dMMR/MSI-H phenotype. Fortunately, active comprehensive treatment, including radical surgery, conveys a survival benefit. In addition, it is necessary to do a comprehensive examination in those who have symptoms as early as possible due to the prognosis of early-stage patients being much better.