Risk factors for the development of metachronous bone metastasis in colorectal cancer patients after curative resection

Utilizing machine learning algorithms for predicting risk factors for bone metastasis from right-sided colon carcinoma after complete mesocolic excision: a 10-year retrospective multicenter study

BACKGROUND

Bone metastasis (BM) occurs when colon cancer cells disseminate from the primary tumor site to the skeletal system via the bloodstream or lymphatic system. The emergence of such bone metastases typically heralds a significantly poor prognosis for the patient. This study's primary aim is to develop a machine learning model to identify patients at elevated risk of bone metastasis among those with right-sided colon cancer undergoing complete mesocolonectomy (CME).

PATIENTS AND METHODS

The study cohort comprised 1,151 individuals diagnosed with right-sided colon cancer, with a subset of 73 patients presenting with bone metastases originating from the colon. We used univariate and multivariate regression analyses as well as four machine learning algorithms to screen variables for 38 characteristic variables such as patient demographic characteristics and surgical information. The study employed four distinct machine learning algorithms, namely, extreme gradient boosting (XGBoost), random forest (RF), support vector machine (SVM), and k-nearest neighbor algorithm (KNN), to develop the predictive model. Additionally, the model was assessed using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA), while Shapley additive explanation (SHAP) was utilized to visualize and analyze the model.

RESULTS

The XGBoost algorithm performed the best performance among the four prediction models. In the training set, the XGBoost algorithm had an area under curve (AUC) value of 0.973 (0.953-0.994), an accuracy of 0.925 (0.913-0.936), a sensitivity of 0.921 (0.902-0.940), and a specificity of 0.908 (0.894-0.922). In the validation set, the XGBoost algorithm had an AUC value of 0.922 (0.833-0.995), an accuracy of 0.908 (0.889-0.926), a sensitivity of 0.924 (0.873-0.975), and a specificity of 0.883 (0.810-0.956). Furthermore, the AUC value of 0.83 for the external validation set suggests that the XGBoost prediction model possesses strong extrapolation capabilities. The results of SHAP analysis identified alkaline phosphatase (ALP) levels, tumor size, invasion depth, lymph node metastasis, lung metastasis, and postoperative neutrophil-to-lymphocyte ratio (NLR) levels as significant risk factors for BM from right-sided colon cancer subsequent to CME.

CONCLUSION

The prediction model for BM from right-sided colon cancer developed using the XGBoost machine learning algorithm in this study is both highly precise and clinically valuable.

A systematic review of population-based studies on metachronous metastases of colorectal cancer

BACKGROUND

The occurrence of metachronous metastases (MM) of colorectal (CRC), colon (CC), and rectal (RC) cancer of population-based studies has not been compiled in a systematic review previously.

METHODS

MEDLINE, Embase, and Cochrane Library were searched for primary studies of any design from inception until January 2021 and updated in August 2023 (CRD42021261648). The PRISMA guidelines were adopted, and the Newcastle-Ottawa Quality Assessment Scale used for risk of bias assessment. Outcomes on overall and organ-specific MM were extracted. A narrative analysis followed.

RESULTS

Out of 2143 unique hits, 162 publications were read in full-text and 37 population-based cohort studies published in 1981-2022 were included. Ten studies adopted time-dependent analyses; eight were registry-based and seven had a low risk of bias. Three studies reported 5-year recurrence rate of MM overall of stages I-III; for CRC, it was 20.5%, for CC, it was 18% and 25.6%, and for RC, it was 23%. Four studies reported 5-year recurrence rate of organ-specific MM of stages I-III-for CRC, it was 2.2% and 5.5% for peritoneal metastases and 5.8% for lung metastases and for CC 4.5% for peritoneal metastases. Twenty-seven studies reported proportions of patients diagnosed with MM, but data on the length of follow-up was incomplete and varied widely. Proportions of patients with CRC stages I-III that developed MM overall was 14.4%-26.1% in 10 studies. In relation to the enrollment period, a downward trend may be discernible.

CONCLUSION

Studies adopting a more appropriate analysis were highly heterogeneous, whereas uncertain data of partly inadequate studies may indicate that MM are overall declining.

Construction and validation of machine learning models for predicting distant metastases in newly diagnosed colorectal cancer patients: A large-scale and real-world cohort study

BACKGROUND

More accurate prediction of distant metastases (DM) in patients with colorectal cancer (CRC) would optimize individualized treatment and follow-up strategies. Multiple prediction models based on machine learning have been developed to assess the likelihood of developing DM.

METHODS

Clinicopathological features of patients with CRC were obtained from the National Cancer Center (NCC, China) and the Surveillance, Epidemiology, and End Results (SEER) database. The algorithms used to create the prediction models included random forest (RF), logistic regression, extreme gradient boosting, deep neural networks, and the K-Nearest Neighbor machine. The prediction models' performances were evaluated using receiver operating characteristic (ROC) curves.

RESULTS

In total, 200,958 patients, 3241 from NCC and 197,717 CRC from SEER were identified, of whom 21,736 (10.8%) developed DM. The machine-learning-based prediction models for DM were constructed with 12 features remaining after iterative filtering. The RF model performed the best, with areas under the ROC curve of 0.843, 0.793, and 0.806, respectively, on the training, test, and external validation sets. For the risk stratification analysis, the patients were separated into high-, middle-, and low-risk groups according to their risk scores. Patients in the high-risk group had the highest incidence of DM and the worst prognosis. Surgery, chemotherapy, and radiotherapy could significantly improve the prognosis of the high-risk and middle-risk groups, whereas the low-risk group only benefited from surgery and chemotherapy.

CONCLUSION

The RF-based model accurately predicted the likelihood of DM and identified patients with CRC in the high-risk group, providing guidance for personalized clinical decision-making.

Prognostic Factors of Bone Metastases From Colorectal Cancer in the Era of Targeted Therapy

BACKGROUND

Various prognostic factors have been reported for bone metastases from different primary tumor sites. However, bone metastases from colorectal cancer are very rare, and the prognostic factors have not been investigated in detail.

OBJECTIVE

This study aimed to identify prognostic factors of bone metastases from colorectal cancer.

DESIGN

This is a retrospective cohort study using data from a prospectively collected database.

SETTINGS

This study was conducted at a single tertiary care cancer center in Japan.

PATIENTS

Patients who developed bone metastases from colorectal cancer during the study period among all patients who received initial treatment for colorectal cancer at our hospital between 2005 and 2016 (n = 4538) were included.

MAIN OUTCOME MEASURES

Overall survival after diagnosis of bone metastases from colorectal cancer was the main outcome measure.

RESULTS

Ninety-four patients developed bone metastases during the study period. The 5-year overall survival rate was 11.0%. Multivariable analysis identified the following independent risk factors associated with poor prognosis: ≥70 years of age at diagnosis of bone metastases (HR, 2.48; 95% CI, 1.24-4.95; p < 0.01), curative surgery not performed as initial treatment (HR, 2.54; 95% CI, 1.24-5.19; p = 0.01), multiple bone metastases (HR, 2.44; 95% CI, 1.30-4.57; p < 0.01), albumin level <3.7 g/dL (HR, 3.80; 95% CI, 1.95-7.39; p < 0.01), CEA ≥30 ng/mL (HR, 1.94; 95% CI, 1.09-3.46; p = 0.02), and less than 3 chemotherapy options remaining at diagnosis of bone metastases (HR, 2.83; 95% CI, 1.51-5.30; p < 0.01). The median survival times for patients with 0-2, 3, and 4-6 risk factors were 25.0, 8.8, and 4.3 months, respectively.

LIMITATIONS

The main limitation is the single-center, retrospective design of this study.

CONCLUSIONS

Our results may facilitate multidisciplinary decision-making in patients with bone metastases from colorectal cancer. See Video Abstract at http://links.lww.com/DCR/B930 .

FACTORES PRONSTICOS DE LAS METSTASIS SEAS DEL CNCER COLORRECTAL EN LA ERA DE LA TERAPIA DIRIGIDA

ANTECEDENTES:Se han reportado varios factores pronósticos para las metástasis óseas de diferentes sitios de tumores primarios. Sin embargo, las metástasis óseas del cáncer colorrectal son muy raras y los factores pronósticos no se han investigado en detalle.OBJETIVO:Identificar los factores pronósticos de las metástasis óseas del cáncer colorrectal.DISEÑO:Estudio de cohorte retrospectivo utilizando datos de una base de datos recolectada prospectivamente.ENTORNO CLINICO:Un solo centro oncológico de atención terciaria en Japón.PACIENTES:Se seleccionaron pacientes que desarrollaron metástasis óseas de cáncer colorrectal durante el período de estudio entre todos los pacientes que recibieron tratamiento inicial para el cáncer colorrectal en nuestro hospital entre 2005 y 2016 (n = 4538).MEDIDA DE RESULTADO PRINCIPAL:Supervivencia general después del diagnóstico de metástasis óseas por cáncer colorrectal.RESULTADOS:Noventa y cuatro pacientes desarrollaron metástasis óseas, lo que representa el 2,0% de todos los pacientes con cáncer colorrectal que comenzaron el tratamiento durante el período de estudio. La tasa de supervivencia global a 5 años fue del 11,0 %. El análisis multivariable identificó los siguientes factores de riesgo independientes asociados con mal pronóstico: edad ≥70 años al momento del diagnóstico de metástasis óseas (hazard ratio 2,48, CI del 95 % 1,24-4,95, p < 0,01), cirugía curativa no realizada como tratamiento inicial (hazard ratio 2,54, CI 95 % 1,24-5,19, p = 0,01), metástasis óseas múltiples (hazard ratio 2,44, CI del 95 % 1,30-4,57, p < 0,01), nivel de albúmina <3,7 g/dL (hazard ratio 3,80, CI del 95 % 1,95 -7,39, p < 0,01), antígeno carcinoembrionario ≥30 ng/mL (hazard ratio 1,94, CI del 95 % 1,09-3,46, p = 0,02) y menos de 3 opciones de quimioterapia restantes al momento del diagnóstico de metástasis óseas (hazard ratio 2,83, 95 % CI 1,51-5,30, p < 0,01). La mediana de los tiempos de supervivencia para los pacientes con 0-2, 3 y 4-6 factores de riesgo fue de 25,0, 8,8 y 4,3 meses, respectivamente.LIMITACIONES:Diseño retrospectivo de un solo centro.CONCLUSIÓN:Nuestros resultados pueden facilitar la toma de decisiones multidisciplinares en pacientes con metástasis óseas de cáncer colorrectal. Consulte Video Resumen en http://links.lww.com/DCR/B930 . (Traducción- Dr. Francisco M. Abarca-Rendon ).

Development and validation of risk and prognostic nomograms for bone metastases in Chinese advanced colorectal cancer patients

Background

Bone metastases (BM) from colorectal cancer (CRC) are often accompanied by extraosseous metastases, resulting in a dismal prognosis. The present study aimed to determine the risk factors for BM in metastatic CRC (mCRC) and the prognostic factors for CRC patients with BM.

Methods

The study was based on a training cohort of 214 mCRC patients (of which, 101 patients had BM) from our center, and a validation cohort of 511 mCRC patients (of which, 173 patients had BM) from another institute. Risk and prognostic nomograms for BM were developed using univariate and multivariate analyses. The goodness of fit, discrimination, and calibration performance of the nomograms were assessed by R², concordance statistics (C-statistics), and the calibration curve. The results were internally validated using bootstrap resampling in the training cohort, and externally validated in the validation cohort.

Results

The novel BM risk nomogram comprised seven variables [degree of tumor differentiation, N-stage, serum alkaline phosphatase (ALP), lactate dehydrogenase (LDH), carcinoembryonic antigen (CEA), liver metastasis, and lung metastasis]. It showed good performance, with an R² of 0.447 and a C-statistic of 0.846 [95% confidence interval (CI), 0.793 to 0.898] in the training cohort, and an R² of 0.325 and a C-statistic of 0.792 (95% CI, 0.750 to 0.834) in the validation cohort. The optimal cutoff value to identify individuals at low or high risk was 56% probability, with a sensitivity of 71.3% and a specificity of 89.4%. The prognostic nomogram included five factors (tumor differentiation, number of extra-BM organs, number of BM lesions, ALP, and LDH), and had an R² of 0.284 and a C-statistic of 0.723 (95% CI, 0.657 to 0.789) in the training set. This nomogram was externally validated in the validation cohort, with an R² of 0.182 and a C-statistic of 0.682 (95% CI, 0.638 to 0.726).

Conclusions

The developed and validated risk and prognostic nomograms showed good performance for predicting the occurrence of BM in mCRC as well as the prognosis of CRC patients with BM. The risk nomogram can be used as a cost-effective preliminary screening tool prior to bone scanning.

Blood biomarkers of bone metastasis in digestive tract malignant tumors

Aim: To evaluate the role of clinical features and blood markers in patients with malignant digestive tract tumors bone metastasis. Materials & methods: A total of 267 patients were included in this trial. Age, gender, primary tumor site, metastatic sites, T/N stage, high-density lipoprotein, low-density lipoprotein, total cholesterol, triglycerides, alkaline phosphatase, LDH, Ca levels, platelet, neutrophils to absolute value of lymphocytes (NLR), ratio of platelets to absolute values of lymphocytes (PLR) were analyzed. Results: T stage, lymph node metastasis, N stage and liver and lung metastasis were independent risk factors. LDH + alkaline phosphatase + NLR + PLR and LDH + NLR, respectively have higher predictive value for bone metastasis compared with patients with early-stage malignant digestive tract tumor and patients with advanced malignant digestive tract tumor without bone metastasis. Conclusion: Some clinical features or blood markers have the potential to detect bone metastasis early to avoid skeletal complications.

Colorectal Cancer and Bone Tissue: Fantastic Relations and Where to Find Them

Colorectal cancer (CRC) is the third most common cancer worldwide. There is a need for the early diagnosis of CRC for a better prognostic outcome. It is, therefore, crucial to understand the CRC pathogenesis in all its aspects. In many cases, one of the main causes of cancer-related deaths is the presence of metastases. In this context, an often overlooked aspect is the metastatic tropism, since CRC, like other cancers, is more prone to metastasize some organs rather than others. Beyond the liver and lung, and differently from other types of cancers, a not usual site of CRC metastases is the bone. However, it may assume a crucial role in the development and the outcome of the disease. Therefore, this review aims to discuss the complex relations between bone markers and CRC pathogenesis, suggesting the use of these molecules as potential targets for therapeutic purposes. Different osteogenic molecules, some of whom are growth factors and are implicated in the different osteogenic pathways, have been proved to also be involved in CRC progression. Some of them are oncogenes, while others oncosuppressors, and in a future perspective, some of them may represent new potential CRC biomarkers.

A clinical model to predict the risk of synchronous bone metastasis in newly diagnosed colorectal cancer: a population-based study

BACKGROUND

The early detection of synchronous bone metastasis (BM) in newly diagnosed colorectal cancer (CRC) affects its initial management and prognosis. A clinical model to individually predict the risk of developing BM would be attractive in current clinical practice.

METHODS

A total of 55,869 CRC patients were identified from Surveillance, Epidemiology, and End Results (SEER) database, of whom 317 patients were diagnosed with synchronous BM. Risk factors for BM in CRC patients was identified using multivariable logistic regression. A weighted scoring system was built with beta-coefficients (P < 0.05). A random sample of 75% of the CRC patients was used to establish the risk model, and the remaining 25% was used to validate its accuracy of this model. The performance of risk model was estimated by receiver operating curve (ROC) analysis.

RESULTS

The risk model consisted of 8 risk factors including rectal cancer, poorly-undifferentiation, signet-ring cell carcinoma, CEA positive, lymph node metastasis, brain metastasis, liver metastasis and lung metastasis. The areas under the receiver operating curve (AUROC) were 0.903 and 0.889 in the development and validation cohort. Patients with scores from 0 to 4 points had about 0.1% risk of developing BM, and the risk increased to about 30% in patients with scores ≥15 points.

CONCLUSIONS

This clinical risk model is accurate enough to identify the CRC patients with high risk of synchronous BM and to further provide more individualized clinical decision.

Preoperative Carcinoembryonic Antigen as a Poor Prognostic Factor in Stage I-III Colorectal Cancer After Curative-Intent Resection: A Propensity Score Matching Analysis

BACKGROUND

Preoperative carcinoembryonic antigen (CEA) has yet to be used as a prognostic or adjuvant chemotherapy factor for colorectal cancer (CRC).

METHODS

This retrospective cohort study included all stage I-III CRC patients with different preoperative serum CEA levels (≤ 5, 5-10, and > 10 ng/ml) at a single center between 1995 and 2010. Propensity score matching was performed in a 1:1 ratio between the two elevated CEA groups (5-10 ng/ml and > 10 ng/ml) and in a 1:2 ratio between the elevated and non-elevated groups (≤ 5 ng/ml), with a caliper of 0.05.

RESULTS

After exclusion and matching, 3857 patients had preoperative CEA levels ≤ 5 ng/ml, 1121 patients had CEA levels between 5 and 10 ng/ml, and 1121 patients had CEA levels > 10 ng/ml. Elevated preoperative CEA showed an increased risk of overall survival (5-10 ng/ml: hazard ratio [HR] 1.376; > 10 ng/ml: HR 1.523; both p < 0.001), cancer-specific survival (5-10 ng/ml: HR 1.404; > 10 ng/ml: HR 1.712; both p < 0.001), and recurrence free interval (5-10 ng/ml: HR 1.190; > 10 ng/ml: HR 1.468; both p < 0.05). Patients with negative lymph node staging (LNs) and CEA > 10 ng/ml, as well as those with positive LNs and CEA ≤ 5 ng/ml, showed similar overall survival (5-year survival: 72% vs. 69%; p = 0.542) and recurrence free intervals (19.9 vs. 21.72 months; p = 0.662).

CONCLUSIONS

A preoperative CEA level can be an independent prognostic factor for stage I-III CRC after curative resection. Patients with negative LNs and preoperative CEA level > 10 ng/ml should be considered for intensive follow-up or adjuvant chemotherapy.

Patterns of bone metastases in newly diagnosed colorectal cancer: a real-world analysis in the SEER database

PURPOSE

To investigate the incidence and the associated factors for bone metastases (BM) development and prognosis in initial colorectal cancer (CRC) with a large sample using Surveillance, Epidemiology, and End Results (SEER) cohort.

METHODS

Primary CRC patients, who were initially diagnosed between 2010 and 2015 in the SEER database, were included to analyze BM incidence and risk factors for BM occurrence. The patients with at least 1-year follow-up were involved to investigate the prognostic factors for BM. Multivariable logistic and proportional hazard regression models were used to investigate the risk factors for BM development and prognosis, respectively.

RESULTS

A total of 212,787 eligible CRC patients were included and 2557 of them were diagnosed with de novo BM (1.20%). Rectal cancer presented significantly higher BM incidence than right and left colon cancer (χ² = 107.64, P < 0.001). T1 stage, poor differentiated grade, and brain metastases were homogeneously associated factors for BM development and BM patients' survival. Male gender, higher N stage, rectal site, elevated carcinoembryonic antigen, and lung and liver metastases were positively associated with BM occurrence. Older age, unmarried status, right colon site, and non-surgery were found to positively correlate with the death risk of CRC patients with BM.

CONCLUSIONS

BM is rare in CRC patients. Homogeneous and heterogeneous factors were found for BM development and BM patients' survival. The risk factors and prognostic factors can be used for BM screening and patient's prognosis estimation.

The risk factors for bone metastases in patients with colorectal cancer

This retrospective analysis aim to evaluate the potential risk factors for bone metastases (BM) in patients who were diagnosed with colorectal cancer (CRC).A total of 2790 patients diagnosed with CRC between January 2006 and December 2016 were collected in this study. All patients were divided into 2 groups, BM and no BM. The associations between biomarkers (including age, gender, histopathological types, alkaline phosphatase (ALP), carcinoembryonic antigen (CEA), cancer antigen 125, and so on), and BM in patients with CRC were analyzed. All the analyses were conducted by SPSS software (version 22.0, SPSS, Chicago, IL).Of all patients, 74 (2.7%) were identified with BM. The level of serum ALP, CEA, and cancer antigen 125 in patients with BM were obviously higher than those without BM (P < .001, P = .005, and P < .001). And the cut-off values of ALP, CEA, and cancer antigen 125 were 85.5 U/L, 6.9 mmol/L, and 16.8 mmol/L, respectively.ALP, CEA, and cancer antigen 125 were identified as the independent risk factors for BM in patients with CRC.

Systematic review: Incidence, risk factors, survival and treatment of bone metastases from colorectal cancer

Background

Bones are not considered a frequent metastatic site in patients with colorectal cancer (CRC). The purpose of the present study was to determine the incidence of bone metastases (BM) in CRC, to identify possible risk factors for BM, survival after BM, and effect of treatment of BM including antiresorptive treatment.

Material and methods

A computer-based literature search was carried out using PubMed and EMBASE.

Results

We included 29 studies. One randomized placebo controlled trial (RCT) study, two autopsy studies, five register studies, and twenty retrospective cohort studies. The studies described different cohorts making direct comparison difficult. Three studies analysed the effect of different treatments for BM including one RCT study.

Conclusion

The incidence of bone metastases was 3–7% in patients with CRC, and it was not possible to detect an increase in incidence over time. The most well established risk factors for BM are rectal cancer, having lymph node invasion at surgery of primary tumor, and lung metastases at any time. Other risk factors such as RAS mutation status have been suggested but results are not conclusive. Survival ranges from 5 to 21 months after diagnosis of BM depending on cohort, with survival of about 8 months in unselected patients. Several variables have been suggested as potential prognostic markers but are all poorly investigated. Treatment of BM is not well investigated, though patients seem to benefit from bisphosphonate treatment with regard to lower risk of skeletal related events. This review highlights the need for new research in the area.

Predictive Nomograms for Synchronous Distant Metastasis in Rectal Cancer

BACKGROUND

Nomograms may be used to quantitatively assess the probability of synchronous distant metastasis. The purpose of this study is to develop predictive nomograms for the presence of synchronous distant metastasis in patients with rectal cancer.

METHODS

A retrospective analysis of the Surveillance Epidemiology and End Results database was performed for cases diagnosed between 2010 and 2014.

RESULTS

Overall, 46,785 patients with rectal cancer (27,773 [59.4%] males, mean age 63.9 ± 13.7 years) were identified, of which 6192 (13.2%) had liver metastasis, 2767 (5.9%) had lung metastasis, and 601 (1.3%) had bone metastasis. Age, sex, race, tumor location, tumor grade, primary tumor size, CEA levels, perineural invasion, T stage, N stage, and liver and lung metastasis were found to be associated with the presence of synchronous distant metastasis and were included in the predictive models. The c-indexes of these models were 0.99 for liver metastasis, 0.99 for lung metastasis, and 1 for bone metastasis.

CONCLUSIONS

Predictive nomograms for the presence of synchronous liver, lung, and bone metastasis were developed and may be used to predict the probability of distant disease in rectal cancer patients.