Data
The SEER database is supported by the National Cancer Institute and collects case data from population-based cancer registries covering approximately 34.6 percent of the U.S. population (14). It contains de-identified data on patient demographics, primary tumor site, tumor morphology, stage at diagnosis, and follow up vital status. For this study we used data of well-differentiated thyroid cancer diagnosed 2004–2010 that were obtained from the November 2017 submission of SEER.
Cases of papillary and follicular thyroid cancer were selected from the SEER 18 databases using the restrictions {ICD-O-3/WHO 2008 = Thyroid} and {Histologic Type ICD-O-3 = 8050, 8260, 8340–8344, 8350, 8450–8460 (for papillary cancer), or Histologic Type ICD-O-3 = 8290, 8330–8335 (for follicular cancer)}. We placed Hurthle cell carcinoma (ICD-O-3 = 8290) into the category of follicular carcinomas, as used in Lim et al. (15). We excluded medullary carcinoma as it is staged differently. Clinically it is often part of genetic syndromes and patients' survival is confounded by its aggressive course. Anaplastic carcinoma is not well-differentiated and uniformly fatal and thus excluded. Furthermore, these 3 types are exceedingly rare and a population-based study on them would be of low utility.
Three datasets were used in this report. Dataset 1, containing 4 prognostic factors, was used to create a prognostic system for thyroid cancer. This system was then compared with the staging system of AJCC. Dataset 2, involving 5 factors, was used to assess the effect of histologic type in creating prognostic systems. Dataset 3, involving 4 factors, was used to explore the optimal cutoff point for age in the prognostic system. Selection of cases, data management, and specifics about factors for each dataset are described as follows.
Dataset 1 SEER started to record derived AJCC T, N, M according to the 6th AJCC Staging Manual in 2004 (16). The AJCC staging systems of thyroid cancer (17–19) contained further stratified categories (T4a and T4b for T and N1a and N1b for N) that were not available in SEER until 2004. Therefore, dataset 1 contained only cases with diagnosis years from 2004 to 2010 to include these categories for catching up with the latest updates in AJCC, and to ensure a 5-year follow-up through 2015 which was the most recent year before which all case-level data were available in SEER. SEER cause-specific death classification variable (20) was used to capture deaths related to thyroid cancer. Survival time was measured by months. The factors in dataset 1 included tumor size (T), regional lymph nodes (N), status of distant metastasis (M), and age (A). The definition of T, N, and M was from Adjusted AJCC 6th ed. T, N, M, and Stage in SEER (16). Age in dataset 1 was treated as a binary variable and contained two categories: A1 (0–54), and A2 (55+). The detailed definition of categories/levels of each factor in dataset 1 was provided in . We excluded patients with a missing or unknown value on any of the following variables: T, N, M, A, survival time, and SEER cause-specific death classification variable. Specifically, we discarded 26 patients with “T4NOS,” 2894 patients with unknown values of T, 2110 patients with “N1NOS,” 1982 patients with unknown values of N, 1,840 patients with unknown values of M, 4 patents with unknown age, 226 patients with unknown survival time, 131 patients with “Dead (missing/unknown COD)” and 8011 patients with “N/A not first tumor.” We note that patients with an unknown value of one variable are more likely to have unknown values on several other variables.
If you are looking for more details, kindly visit tnma.
Table 1
FactorsLevelsDefinitionsPrimary tumorT0No evidence of primary tumorT1Tumor 2 cm or less in greatest dimension limited to the thyroidT2Tumor more than 2 cm but not more than 4 cm in greatest dimension limited to the thyroidT3Tumor more than 4 cm in greatest dimension limited to the thyroid or any tumor with minimal extrathyroid extension (e.g., extension to sternothyroid muscle or perithyroid soft tissues)T4aTumor of any size extending beyond the thyroid capsule to invade subcutaneous soft tissues, larynx, trachea, esophagus, or recurrent laryngeal nerveT4bTumor invades prevertebral fascia or encases carotid artery or mediastinal vesselsRegional nodes positiveN0No regional lymph node metastasisN1aMetastasis to Level VI (pretracheal, paratracheal, and prelaryngeal/Delphian lymph nodes)N1bMetastasis to unilateral, bilateral, or contralateral cervical or superior mediastinal lymph nodesMetastasisM0No distant metastasisM1Distant metastasisAgeA10 ≤ Age < 55A255 ≤ AgeOpen in a separate window
In creating prognostic systems based on dataset 1, our approach applied to combinations instead of individual patients. A combination of prognostic factors is a subset of the data that corresponds to one level of each selected factor. A combination describes certain characteristics of its patients. For example, T1 and N0 produce a combination, denoted by T1N0, which represents a subset of patients whose tumor size is T1 and lymph nodes positive is N0. As in T1N0, we use the notations of levels of factors to denote combinations in this report.
To optimize robustness of statistical techniques, we only kept combinations (in terms of T, N, M, A) each containing at least 25 patients in dataset 1. This left out 33 “rare” combinations (321 cases). Note that 68 cases had T0 and they were excluded since all combinations involving T0 contained fewer than 25 patients. The final dataset 1 consisted of 39 combinations (51,291 cases with a median follow up 90 months).
Dataset 2 Dataset 2 was derived from dataset 1 by treating the histology (H) as an additional prognostic factor. Two levels were used for histologic type: H1 (follicular) and H2 (papillary). To optimize robustness of statistical techniques, we only kept combinations (in terms of T, N, M, A, H) each containing at least 25 patients. This left out 268 cases from dataset 1. The final dataset 2 consisted of 44 combinations (51,023 cases with a median follow up 90 months). This is the largest dataset that contains combinations (in terms of T, N, M, A, and H) each containing at least 25 patients with diagnosis years from 2004 to 2010.
Dataset 3 Dataset 3 was also derived from dataset 1 due to consideration of three cutoffs of age 45, 55, and 65. Both 45 and 55 have been used in recent editions of AJCC, and 65 was considered because of its general use in the literature for stratifying young and old patients. We required that each combination from T, N, M, and any cutoff contain at least 25 patients. This left out 186 cases from dataset 1. The final dataset contains 51,105 cases (with a median follow up 90 months), which is the largest dataset that contains combinations (in terms of T, N, M, A with any of the three cutoffs) each containing at least 25 patients with diagnosis years from 2004 to 2010.
Additional resources:How do you bench test a water meter?
Hangzhou Tnma MeasureTnma Measure Technology Co.,Ltd's Post
Rising consumer expectations continue to highlight the need for robust customer training. To help customers get the most value out of products and services, companies rely on education software to streamline user training.
But why should your team tap into customer training software? Here are the top five benefits.
Read more: https://absorbl.ms/3q2kL4c
#Learning #LMS #CustomerTraining #TrainingandDevelopment
Creating Prognostic Systems for Well-Differentiated Thyroid ...
Data
The SEER database is supported by the National Cancer Institute and collects case data from population-based cancer registries covering approximately 34.6 percent of the U.S. population (14). It contains de-identified data on patient demographics, primary tumor site, tumor morphology, stage at diagnosis, and follow up vital status. For this study we used data of well-differentiated thyroid cancer diagnosed 2004–2010 that were obtained from the November 2017 submission of SEER.
Cases of papillary and follicular thyroid cancer were selected from the SEER 18 databases using the restrictions {ICD-O-3/WHO 2008 = Thyroid} and {Histologic Type ICD-O-3 = 8050, 8260, 8340–8344, 8350, 8450–8460 (for papillary cancer), or Histologic Type ICD-O-3 = 8290, 8330–8335 (for follicular cancer)}. We placed Hurthle cell carcinoma (ICD-O-3 = 8290) into the category of follicular carcinomas, as used in Lim et al. (15). We excluded medullary carcinoma as it is staged differently. Clinically it is often part of genetic syndromes and patients' survival is confounded by its aggressive course. Anaplastic carcinoma is not well-differentiated and uniformly fatal and thus excluded. Furthermore, these 3 types are exceedingly rare and a population-based study on them would be of low utility.
Three datasets were used in this report. Dataset 1, containing 4 prognostic factors, was used to create a prognostic system for thyroid cancer. This system was then compared with the staging system of AJCC. Dataset 2, involving 5 factors, was used to assess the effect of histologic type in creating prognostic systems. Dataset 3, involving 4 factors, was used to explore the optimal cutoff point for age in the prognostic system. Selection of cases, data management, and specifics about factors for each dataset are described as follows.
Dataset 1 SEER started to record derived AJCC T, N, M according to the 6th AJCC Staging Manual in 2004 (16). The AJCC staging systems of thyroid cancer (17–19) contained further stratified categories (T4a and T4b for T and N1a and N1b for N) that were not available in SEER until 2004. Therefore, dataset 1 contained only cases with diagnosis years from 2004 to 2010 to include these categories for catching up with the latest updates in AJCC, and to ensure a 5-year follow-up through 2015 which was the most recent year before which all case-level data were available in SEER. SEER cause-specific death classification variable (20) was used to capture deaths related to thyroid cancer. Survival time was measured by months. The factors in dataset 1 included tumor size (T), regional lymph nodes (N), status of distant metastasis (M), and age (A). The definition of T, N, and M was from Adjusted AJCC 6th ed. T, N, M, and Stage in SEER (16). Age in dataset 1 was treated as a binary variable and contained two categories: A1 (0–54), and A2 (55+). The detailed definition of categories/levels of each factor in dataset 1 was provided in . We excluded patients with a missing or unknown value on any of the following variables: T, N, M, A, survival time, and SEER cause-specific death classification variable. Specifically, we discarded 26 patients with “T4NOS,” 2894 patients with unknown values of T, 2110 patients with “N1NOS,” 1982 patients with unknown values of N, 1,840 patients with unknown values of M, 4 patents with unknown age, 226 patients with unknown survival time, 131 patients with “Dead (missing/unknown COD)” and 8011 patients with “N/A not first tumor.” We note that patients with an unknown value of one variable are more likely to have unknown values on several other variables.
Table 1
FactorsLevelsDefinitionsPrimary tumorT0No evidence of primary tumorT1Tumor 2 cm or less in greatest dimension limited to the thyroidT2Tumor more than 2 cm but not more than 4 cm in greatest dimension limited to the thyroidT3Tumor more than 4 cm in greatest dimension limited to the thyroid or any tumor with minimal extrathyroid extension (e.g., extension to sternothyroid muscle or perithyroid soft tissues)T4aTumor of any size extending beyond the thyroid capsule to invade subcutaneous soft tissues, larynx, trachea, esophagus, or recurrent laryngeal nerveT4bTumor invades prevertebral fascia or encases carotid artery or mediastinal vesselsRegional nodes positiveN0No regional lymph node metastasisN1aMetastasis to Level VI (pretracheal, paratracheal, and prelaryngeal/Delphian lymph nodes)N1bMetastasis to unilateral, bilateral, or contralateral cervical or superior mediastinal lymph nodesMetastasisM0No distant metastasisM1Distant metastasisAgeA10 ≤ Age < 55A255 ≤ AgeOpen in a separate window
In creating prognostic systems based on dataset 1, our approach applied to combinations instead of individual patients. A combination of prognostic factors is a subset of the data that corresponds to one level of each selected factor. A combination describes certain characteristics of its patients. For example, T1 and N0 produce a combination, denoted by T1N0, which represents a subset of patients whose tumor size is T1 and lymph nodes positive is N0. As in T1N0, we use the notations of levels of factors to denote combinations in this report.
To optimize robustness of statistical techniques, we only kept combinations (in terms of T, N, M, A) each containing at least 25 patients in dataset 1. This left out 33 “rare” combinations (321 cases). Note that 68 cases had T0 and they were excluded since all combinations involving T0 contained fewer than 25 patients. The final dataset 1 consisted of 39 combinations (51,291 cases with a median follow up 90 months).
Dataset 2 Dataset 2 was derived from dataset 1 by treating the histology (H) as an additional prognostic factor. Two levels were used for histologic type: H1 (follicular) and H2 (papillary). To optimize robustness of statistical techniques, we only kept combinations (in terms of T, N, M, A, H) each containing at least 25 patients. This left out 268 cases from dataset 1. The final dataset 2 consisted of 44 combinations (51,023 cases with a median follow up 90 months). This is the largest dataset that contains combinations (in terms of T, N, M, A, and H) each containing at least 25 patients with diagnosis years from 2004 to 2010.
Dataset 3 Dataset 3 was also derived from dataset 1 due to consideration of three cutoffs of age 45, 55, and 65. Both 45 and 55 have been used in recent editions of AJCC, and 65 was considered because of its general use in the literature for stratifying young and old patients. We required that each combination from T, N, M, and any cutoff contain at least 25 patients. This left out 186 cases from dataset 1. The final dataset contains 51,105 cases (with a median follow up 90 months), which is the largest dataset that contains combinations (in terms of T, N, M, A with any of the three cutoffs) each containing at least 25 patients with diagnosis years from 2004 to 2010.
5
0
0
Comments
All Comments ( 0 )