Abstract
Objective: To characterize the clinicopathologic features and to investigate the prognostic nomograms for overall survival (OS) and cancer-specific survival (CSS) in patients with Hepatic malignant vascular tumors (HMVT). Method: Patients diagnosed with HMVT between 1973 and 2015 were screened from the Surveillance, Epidemiology, and End Results (SEER) database. The Kaplan-Meier (KM) was used for survival analysis. The univariate and multivariate Cox analyses were performed to identify independent predictors. Furthermore, the prognostic nomograms were established and evaluated. Results: A total of 510 HMVT patients were collected, and randomly divided into HMVT-training (N=308) and validation cohort (N=202) groups. The 3- and 5-year OS for overall HMVT were 21.3% and 19.8%, and the corresponding CSS was 29.8% and 27.7% respectively. Age at diagnosis, grade, tumor size, and histological type were identified as prognostic factors for OS and CSS in patients with HMVT. However, sex was just for predicting CSS, and T stage was only an indicator of OS. These factors were further utilized to construct the nomograms for OS and CSS in the HMVT-training cohort showing credible performance with the C-index of 0.763 and 0.762, respectively. Moreover, the AUC value for 1-, 3-, 5-year OS was 0.873, 0.905 and 0.898, and the corresponding value for CSS was 0.808, 0.794 and 0.788 respectively. Additionally, the calibration curves demonstrated a favorable agreement between the predicted and actual 1-, 3- and 5-year survival rates both in the training and validated cohorts. Conclusion: This was the largest population-based study to describe the clinicopathologic characteristics in patients with HMVT. Moreover, we established and validated prognostic nomograms that indicated an accurate prediction for 1-, 3- and 5-year of OS and CSS.
Keywords: Hepatic malignant vascular tumors, prognostic factor, nomogram, overall survival, cancer-specific survival, SEER database
Introduction
Malignant vascular tumors (MVT) are rare types of tumors and can occur in various organs, including the liver, lung, bone, pleura, spleen, and lymph nodes et al. To date, the number of reported cases of hepatic malignant vascular tumors (HMVT) is limited. Angiosarcoma (AS), hemangioendothelioma (HE), epithelioid hemangioendothelioma (EHE), and hemangiopericytoma (HP) are the most common histological types of HMVT [1-3]. With regard to primary liver tumors such as hepatocellular carcinoma and cholangiocarcinoma, standard criteria have been established to determine the curative or palliative outcomes of surgical treatment or systematic therapies. However, because of the extremely rare incidence of these types, little is known about the pathogenesis and progression of HMVT both in clinical trials and laboratory research, which hindered the researchers from elaborating the pathological mechanisms and determining the prognostic factors that are closely correlated with patients’ survival [1,4,5].
AS is a subtype of soft tissue sarcoma and has aggressive and malignant features derived from the endothelial cell tumors in the vessels or lymph nodes [2]. Although the development of several vascular-targeted treatments has attracted much interest in identifying the molecular pathogenesis and clinical investigations, currently, the feasible treatments for AS are still limited and the prognosis is poor. The most common primary sites of AS included the neck (37%), head (52%), extremities (15.3%), trunk (9.5%), and liver (6.0%). Meanwhile, AS accounts for one-third of all cases of HMVT and is more generally diagnosed in patients aged 50-70 years [6]. Histologically, it is composed of vascular channels that are lined by variably atypical endothelial cells with large nucleoli, nuclei, and incremental mitosis [7,8]. In 1949, Stout first described HE as neoplasms arising from pericytes with neoplastic cellular infiltration in the sinusoids and intrahepatic veins [9]. HE less commonly occurs in the liver compared to AS, preferably develops in patients aged 40-50 years, and accounts for 2% of soft tissue tumors. However, the overall malignancy of HE is weaker than that of AS. Patients with HE have a better prognosis with a 5-year overall survival (OS) of 60% [2,10]. In such cases, epithelioid-type tumors would develop in the infiltrating cells, leading to a subtype classified as EHE. To date, only a few studies have reported the risk factors associated with OS in patients with HE and EHE, such as tumor size, invasiveness, and foci of hemorrhage or necrosis [5,11,12]. The HP is an extremely rare type of tumor, accounting for less than 1% of all vascular neoplasms, and develops more often in women than in men. It is usually detected in several body sites, including the lower limbs, abdominal and retroperitoneal cavity, and liver, with features of single or multiple cases [1,13,14]. Pericytes are myofibroblast-like cells that wrap around the capillaries and venules, and their malignant transformation finally results in HP [15,16]. Hitherto, aside from a few sporadic cases, knowledge about the natural pathogenesis, etiologic causes, and prognostic factors of HP is extremely scarce.
To date, local disease control and improvement of survival probability in patients with HMVT remain a challenge. Only a few studies systematically described the clinical features or reported the significant morbidity and complications associated with surgical treatment and other therapeutic approaches. Therefore, further investigations are warranted to determine whether a uniform therapeutic strategy should be applied to all patients regardless of histopathology or whether an individualized treatment should be adopted. Moreover, the potential prognostic factors associated with survival are deserved to clarify.
This retrospective analysis was performed on patients with HMVT based on the data from the Surveillance, Epidemiology, and End Results (SEER) database, which is supported by the United States (US) National Cancer Institute’s surveillance program. It is a real-world cancer clinical document database that has been widely utilized for cancer clinical studies, especially for evaluating the incidence of rare tumor types. A total of 510 HMVT cases were retrieved from the SEER database between 1973 and 2015. In the current, this is the largest retrospective study to perform an in-depth analysis of the clinicopathological traits and establish the prognostic nomograms models for predicting 1-, 3- and 5-year OS and cancer-specific survival (CSS) in patients with HMVT, respectively.
Materials and method
Data source and acquisition
The patient data between 1973 and 2015 were extracted from the SEER database, the world’s largest publicly accessible cancer registry, supported by the US National Cancer Institute. Approval from the institutional review board is not required when using SEER data due to the absence of patient identifiers. SEER* Stat (version 8.3.6) was used to screen patients diagnosed with HMVT according to the International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3rd). All selected patients met the inclusion criteria based on the corresponding morphological codes, including 9120 (angiosarcoma), 9130 (hemangioendothelioma), 9133 (epithelioid hemangioendothelioma), and 9150 (hemangiopericytoma), all of which were combined with site code C22.0 (liver) and the sequence number of one primary only.
The demographic and clinicopathological characteristics of each patient were carefully collected and classified by age, sex, race, year of diagnosis, alpha-fetoprotein protein (AFP) level, degree of fibrosis, tumor number, tumor size, tumor-node-metastasis (TNM) stage, histologic types, radiation, chemotherapy, and surgery. Simultaneously, the 4-grade system in the SEER database was used to define the level of tumor differentiation: well-differentiated (grade I), moderately differentiated (grade II), poorly differentiated (grade III), and undifferentiated (grade IV). Additionally, the SEER database also provided straightforward reports of OS, defined as the period from diagnosis to death from any cause or last follow-up, and CSS, defined as the overall time from diagnosis to death specifically due to cancers.
Establishment and validation of the nomogram
The univariate and multivariate Cox regression analyses were conducted to determine the independent prognostic factors correlated with survival probability. Then the predictive nomograms for OS and CSS in patients with HMVT were established based on the screened risk factors. To further evaluate the predictive performance of the nomogram models in both the training and validation cohorts, the concordance index (C-index) and the area under the curve (AUC) of time-dependent receiver operating characteristic (ROC) were conducted. Generally, the value of C-index is 0.7 or higher indicating a credible prediction. The range of AUC value is from 0 to 1, and a model was thought of as a poor, reliable, or excellent performance with the AUC value interval of 0.5 to 0.6, 0.6 to 0.7, or more than 0.7, respectively. Besides, the calibration curve analysis was performed to evaluate the consistency between the predicted and practical survival times. The whole assessment process of the nomogram was operated in R software (4.0.3 version) (https://www.r-project.org/) by R packages of rms, survival, and survminer.
Risk score calculation and risk group classification
The risk score formula was constructed to quantitatively calculate the risk of inferior survival probability for each patient according to the coefficients of the independent clinicopathological risk factors determined by multivariate Cox regression analysis. The “survival” R package was used for the process. We eventually divided the patients into low-risk and high-risk groups based on the median risk scores. The OS and CSS for patients in different risk stratification classified by clinical-pathological factors were further evaluated.
Statistical analysis
Statistical analyses were carried out using the universal statistical SPSS software (version 22.0) from BM Corp, Armonk, NY, USA, and a P-value of <0.05 was considered statistically significant. Categorical and continuous variables were analyzed using the Pearson chi-square test and 2-sample t-test, respectively. The Kaplan-Meier (KM) method was used to calculate the survival rates and median survival time, while the log-rank test was applied to thoroughly evaluate the deviations. All variables with a P-value of <0.05 in univariate analysis were selected for the multivariate proportional hazard model to identify independent predictors for OS and CSS. Hazard ratios (HRs) and 95% confidence intervals (CIs) were also calculated.
Results
Demographic and clinicopathological characteristics of patients
Between 1973 and 2015, a total of 510 patients with HMVT were identified from the SEER database, of whom 350 had AS, 37 had HE, 114 had EHE, and 9 had HP. These patients were randomly divided into the training cohort (N=308) and the validation cohort (N=202). The demographic and clinicopathological characteristics of all included patients are summarized in Table 1. The detailed distribution of the age at diagnosis of patients with different types of HMVT ranged from 0 to ≥84 years and was presented in Figure S1. Meanwhile, we used X-title software to determine the cutoff value of age at diagnosis to divide all patients into three groups, namely age <40, 40-59, and ≥60 years (Figure S2). Among patients with HMVT, 69 (13.5%) were aged <40 years, 179 (35.1%) were aged 40-59 years, and 262 (51.4%) were aged 60 years and older. Moreover, the incidence rates of HMVT in men and women showed no obvious difference (45.7% and 54.3%, respectively). The majority of patients with AS (78.6%), HE (86.5%), EHE (80.7%), HP (66.7%), and overall HMVT (79.4%) were white people. With regard to the year of diagnosis for HMVT, 133 patients (26.1%) were diagnosed between 1973 and 1999, 127 patients (24.9%) between 2000 and 2005, 106 patients (20.8%) between 2006 and 2010, and 144 patients (28.2%) between 2011 and 2015. Only 136 (26.6%) and 32 (6.3%) patients with HMVT had data to define AFP level and degree of fibrosis. Approximately 2.6% of AS patients and 4.4% of EHE patients showed elevated AFP levels. Correspondingly, 16 patients (14.6%) with AS and 2 patients (1.8%) with EHE showed severe fibrosis. Among HMVT patients, 89.2% had a single lesion and 12.5% had tumor sizes of ≤5 cm.
Table 1.
Demographics and clinicopathological characteristics of patients with hepatic malignant vascular tumor (HMVT)
| Categories | Number (%) | P value | ||||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| AS | HE | EHE | HP | HMVT | HMVT (Training) | HMVT (Validation) | ||
| (N=350) | (N=37) | (N=114) | (N=9) | (N=510) | (N=308) | (N=202) | ||
| Age, years | <0.001 | |||||||
| <40 | 29 (8.3) | 9 (24.3) | 30 (26.3) | 1 (11.1) | 69 (13.5) | 48 (15.6) | 21 (10.4) | |
| 40-59 | 111 (31.7) | 14 (37.8) | 48 (42.1) | 6 (66.7) | 179 (35.1) | 104 (33.8) | 75 (37.1) | |
| ≥60 | 210 (60) | 14 (37.8) | 36 (31.6) | 2 (22.2) | 262 (51.4) | 156 (50.6) | 106 (52.5) | |
| Race | 0.442 | |||||||
| white | 275 (78.6) | 32 (86.5) | 92 (80.7) | 6 (66.7) | 405 (79.4) | 237 (76.9) | 178 (83.2) | |
| black | 17 (4.9) | 2 (5.4) | 11 (9.6) | 1 (11.1) | 31 (6.1) | 19 (6.2) | 12 (5.9) | |
| other | 58 (16.6) | 3 (8.1) | 11 (9.6) | 2 (22.2) | 74 (14.5) | 52 (16.9) | 22 (10.9) | |
| Sex | <0.001 | |||||||
| female | 126 (36) | 24 (64.9) | 77 (67.5) | 6 (66.7) | 233 (45.7) | 149 (48.4) | 84 (41.6) | |
| male | 224 (64) | 13 (35.1) | 37 (32.5) | 3 (33.3) | 277 (54.3) | 159 (51.6) | 118 (58.4) | |
| Year of diagnosis | 0.017 | |||||||
| 1973-1999 | 95 (27.1) | 20 (54.1) | 15 (13.2) | 3 (33.3) | 133 (26.1) | 76 (24.7) | 57 (28.2) | |
| 2000-2005 | 84 (24) | 6 (16.2) | 35 (30.7) | 2 (22.2) | 127 (24.9) | 83 (26.9) | 44 (21.8) | |
| 2006-2010 | 66 (18.9) | 4 (10.8) | 33 (28.9) | 3 (33.3) | 106 (20.8) | 65 (21.1) | 41 (20.3) | |
| 2010-2015 | 105 (30) | 7 (18.9) | 31 (27.2) | 1 (11.1) | 144 (28.2) | 84 (27.3) | 60 (29.7) | |
| Grade | 0.066 | |||||||
| I | 8 (2.3) | 0 (0.0) | 4 (3.5) | 0 (0.0) | 12 (2.4) | 7 (2.3) | 5 (2.5) | |
| II | 10 (2.9) | 0 (0.0) | 10 (8.8) | 1 (11.1) | 21 (4.1) | 13 (4.2) | 8 (4.0) | |
| III | 33 (9.4) | 0 (0.0) | 4 (3.5) | 1 (11.1) | 38 (7.5) | 23 (7.5) | 15 (7.4) | |
| IV | 35 (10) | 0 (0.0) | 1 (0.9) | 1 (11.1) | 37 (7.3) | 24 (7.8) | 13 (6.4) | |
| unknown | 264 (75.4) | 37 (100.0) | 95 (83.3) | 6 (66.7) | 402 (78.8) | 241 (78.2) | 161 (79.7) | |
| AFP | 0.94 | |||||||
| normal | 86 (24.9) | 5 (13.5) | 29 (25.4) | 2 (22.2) | 122 (23.9) | 75 (24.4) | 47 (23.3) | |
| elevated | 9 (2.6) | 0 (0.0) | 5 (4.4) | 0 (0.0) | 14 (2.7) | 8 (2.6) | 6 (3.0) | |
| unknown | 255 (72.9) | 32 (86.5) | 80 (70.2) | 7 (77.8) | 374 (73.3) | 225 (73.1) | 149 (73.8) | |
| Fibrosis | 0.402 | |||||||
| none to moderate | 7 (2.0) | 0 (0.0) | 6 (5.3) | 1 (11.1) | 14 (2.7) | 11 (3.6) | 3 (1.5) | |
| Severe | 16 (14.6) | 0 (0.0) | 2 (1.8) | 0 (0.0) | 18 (3.5) | 11 (3.6) | 7 (3.5) | |
| unknown | 327 (93.4) | 37 (100.0) | 106 (93.0) | 8 (88.9) | 478 (93.7) | 286 (92.9) | 192 (95.0) | |
| Tumor number | 0.64 | |||||||
| 1 | 309 (88.3) | 36 (97.3) | 103 (90.4) | 7 (77.8) | 455 (89.2) | 270 (87.7) | 185 (91.6) | |
| 2 | 34 (9.7) | 1 (2.7) | 11 (9.6) | 2 (22.2) | 47 (9.2) | 31 (10.1) | 16 (7.9) | |
| ≥3 | 7 (2.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 7 (1.4) | 6 (1.9) | 1 (0.5) | |
| Tumor size | 0.001 | |||||||
| ≤5 cm | 29 (8.3) | 5 (13.5) | 30 (26.3) | 0 (0.0) | 64 (12.5) | 44 (14.3) | 20 (9.9) | |
| >5 cm | 88 (25.1) | 4 (10.8) | 17 (14.9) | 3 (33.3) | 112 (22.0) | 66 (21.4) | 46 (22.8) | |
| unknown | 233 (66.6) | 28 (75.5) | 67 (58.8) | 6 (66.7) | 334 (65.5) | 198 (64.3) | 136 (67.3) | |
| Surgery | 0.048 | |||||||
| Yes | 14 (4) | 4 (10.8) | 0 (0.0) | 1 (11.1) | 19 (3.7) | 8 (2.6) | 11 (5.4) | |
| No | 37 (10.6) | 3 (8.1) | 9 (7.9) | 2 (22.2) | 51 (10.0) | 24 (7.8) | 27 (13.4) | |
| unknown | 299 (85.4) | 30 (81.1) | 105 (92.1) | 6 (66.7) | 440 (86.3) | 276 (89.6) | 164 (81.2) | |
| Radiation | 0.567 | |||||||
| Yes | 12 (3.4) | 0 (0.0) | 2 (1.8) | 0 (0.0) | 14 (2.7) | 8 (2.6) | 6 (3.0) | |
| No | 338 (96.6) | 37 (100.0) | 112 (98.2) | 9 (100.0) | 496 (97.3) | 300 (97.4) | 196 (97.0) | |
| Chemotherapy | 0.369 | |||||||
| yes | 48 (13.7) | 9 (24.3) | 18 (15.8) | 1 (11.1) | 76 (14.9) | 51 (16.6) | 25 (12.4) | |
| No | 302 (86.3) | 28 (75.7) | 96 (84.2) | 8 (88.9) | 434 (85.1) | 257 (83.4) | 177 (87.6) | |
| T stage | 0.521 | |||||||
| T1 | 48 (13.7) | 4 (10.8) | 21 (18.4) | 2 (22.2) | 75 (14.7) | 48 (15.6) | 27 (13.4) | |
| T2 | 24 (6.9) | 4 (10.8) | 21 (18.4) | 0 (0.0) | 49 (9.6) | 29 (9.4) | 20 (9.9) | |
| T3 | 43 (12.3) | 2 (5.4) | 10 (8.8) | 1 (11.1) | 56 (11.0) | 35 (11.4) | 21 (10.4) | |
| T4 | 11 (3.1) | 1 (2.7) | 5 (4.4) | 1 (11.1) | 18 (3.5) | 11 (3.6) | 7 (3.5) | |
| unknown | 224 (64.0) | 26 (70.3) | 57 (50.0) | 5 (55.6) | 312 (61.2) | 185 (60.1) | 127 (62.9) | |
| N stage | 0.051 | |||||||
| N0 | 154 (44.0) | 10 (27.0) | 59 (51.8) | 3 (33.3) | 226 (44.3) | 136 (44.2) | 90 (44.6) | |
| N1 | 8 (2.3) | 0 (0.0) | 9 (7.9) | 1 (11.1) | 18 (3.5) | 11 (3.6) | 7 (3.5) | |
| unknown | 188 (53.7) | 27 (73.0) | 46 (40.4) | 5 (55.6) | 266 (52.2) | 161 (52.3) | 105 (52) | |
| M stage | 0.159 | |||||||
| M0 | 117 (33.4) | 7 (18.9) | 39 (34.2) | 3 (33.3) | 166 (32.5) | 100 (32.5) | 66 (32.7) | |
| M1 | 62 (17.7) | 6 (16.2) | 35 (30.7) | 1 (11.1) | 104 (20.4) | 65 (21.1) | 39 (19.3) | |
| unknown | 171 (48.9) | 24 (64.9) | 40 (35.1) | 5 (55.6) | 240 (47.1) | 143 (46.4) | 97 (48.0) | |
| OS | ||||||||
| 1 year-OS | 51 (12.6) | 17 (45.3) | 83 (71.9) | 4 (44.4) | 155 (28.9) | 94(29.3) | 61 (28.4) | |
| 3 year-OS | 31 (5.9) | 16 (42.5) | 72 (60.5) | 3 (33.3) | 122 (21.3) | 76 (22.6) | 46 (19.4) | |
| 5 year-OS | 31 (5.9) | 15 (39.2) | 67 (54.8) | 3 (33.3) | 116 (19.8) | 71 (20.6) | 45 (18.7) | |
| CSS | ||||||||
| 1 year-CSS | 113 (20.0) | 20 (51.0) | 92 (78.9) | 5 (53.3) | 230 (37.9) | 144 (39.8) | 86 (35.1) | |
| 3 year-CSS | 98 (11.2) | 20 (51.0) | 81 (66.4) | 5 (53.3) | 204 (29.8) | 130 (32.5) | 74 (25.8) | |
| 5 year-CSS | 98 (11.2) | 19 (47.1) | 76 (60.1 | 5 (53.3) | 198 (27.7) | 125 (29.6) | 73 (24.9) | |
AS: angiosarcoma. HE: hemangioendothelioma. EHE: epithelioid hemangioendothelioma. HP: hemangiopericytoma. HMVT: hepatic malignant vascular tumor. TNM: tumor-node-metastasis. OS: overall survival. CSS: cancer-specific survival. Chi-square was used for statistical analysis, P<0.05 was considered statistically significant.
Histologically, there were no HE patients with the information of grade levels. Among the HMVT patients, 402 (78.8%) had an unknown tumor grade, 33 (6.5%) had low-grade tumors (grade I+II), and 75 (14.8%) had high-grade tumors (grade III+IV). Concerning the T stage, 124 (24.3%) patients with HMVT had early T stage (T1+T2), while 74 (14.5%) had advanced T stage (T3+T4). In the early T stage group, 72 (20.6%) patients had AS, while 42 (36.8%) had EHE; in the advanced T stage group, 54 (15.4%) patients had AS, while 15 (13.2%) had EHE. A total of 226 (44.3%) patients with HMVT had N0 stage, while only 3.5% had N1 stages. Likewise, 166 (32.5%) patients had M0 stage, while 104 (20.4%) had M1 stage.
As to the treatments, only 19 (3.7%), 14 (2.7%), 76 (14.9%) patients with HMVT received surgery, radiation, and chemotherapy treatment, separately. While the major population had an unknown treatment record. Furthermore, the OS and CSS were calculated for HMVT and the corresponding histologic types (Figure 1). It was indicated that the 1- and 3-year OS and CSS rates for AS were 12.6% and 5.9%, and 20.0% and 11.2%, respectively. The 5-year OS and CSS rates were 39.2% and 47.1% for HE, 54.8% and 60.1% for EHE, 33.3% and 53.3% for HP, and 19.8% and 27.7% for HMVT, respectively. Additionally, the description of clinicopathologic features of patients in the HMVT-training and -validation group was also analyzed in Table 1.
Figure 1.
The tendency of overall survival (OS) and cancer-specific survival (CSS) for patients with angiosarcoma (OS), hemangioendothelioma (HE), epithelioid hemangioendothelioma (EHE), hemangiopericytoma (HP), and all hepatic malignant vascular tumors (HMVT) respectively using the data from Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2015. A, B: Indicated the OS and CSS for AS, HE, EHE, HP, and HMVT respectively.
Univariate analysis of variables correlated with OS and CSS
At first, the univariate analysis using the log-rank test was performed to identify the potential prognostic factors correlated with OS and CSS in the AS, EHE, and HMVT-all, and HMVT-training cohorts (Table 2). Our tests indicated that age ≥60 years was significantly associated with poorer OS and CSS in the AS (P<0.001 and P=0.003), EHE (P<0.001 and P=0.01), HMVT-all (both P<0.001), and HMVT-training (both P<0.001). Similarly, sex was also a poorer predictor of OS and CSS in the AS, HMVT-all, and HMVT-training cohorts (P=0.010 and P=0.015 for AS, both P<0.001 for HMVT-all, and both P<0.001 for HMVT-training, respectively). Neither race nor AFP level was significantly associated with OS or CSS in any type of HMVT cohorts. Moreover, tumor grade was a significant predictor for predicting OS and CSS in the HMVT-all and -training cohort but just associated with CSS (P=0.0307) in the EHE cohort. Patients with severe fibrosis exhibited lower survival probability in both OS and CSS than those with no to moderate level of fibrosis in the EHE (P=0.022 and P=0.014, respectively) and HMVT-all cohorts (P=0.017 and P=0.006, respectively). The significant difference in CSS brought by tumor numbers was observed in the AS (P<0.001), HMVT-all (P<0.001), and HMVT-training (P=0.004) cohorts. Notably, a smaller tumor size (≤5 cm) was a beneficial predictor for both OS and CSS in the AS, EHE, HMVT-all, and -training cohorts. In the current study, AS patients who underwent surgery showed better OS (P=0.016) and CSS (P=0.062). Similarly, the patients with AS treated by chemotherapy lived a longer time than those who didn’t receive this treatment (P=0.033 for OS). As well, statistical differences were observed in the OS and CSS between patients with early T stage and those with advanced T stage in the EHE (P=0.025 and P=0.016), HMVT-all (P=0.001 and P=0.003), and HMVT-training (P=0.008 and P=0.024) cohorts. Meanwhile, the difference for CSS in patients stratified by N stage was only identified in the EHE cohort with a P-value of 0.036. However, regarding M stages, we did not identify any significant differences in the OS and CSS for HMVT patients with various histological types. Additionally, we found that the histological type of HMVT is a predictive factor associated with survival probability both for OS and CSS.
Table 2.
Univariate analysis of OS and CSS for patients with hepatic malignant vascular tumor (HMVT)
| Categories | AS (N=350) | EHE (N=114) | HMVT-all (N=510) | HMVT-training | ||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| P-value, HR (95% CI) | ||||||||
|
| ||||||||
| OS | CSS | OS | CSS | OS | CSS | OS | CSS | |
| Age | ||||||||
| <40 | reference | reference | reference | reference | ||||
| 40-59 | 0.185, 1.37 (0.86-2.19) | 0.277, 1.32 (0.80-2.18) | 0.378, 1.41 (0.66-3.05) | 0.637, 1.21 (0.55-2.68) | 0.005, 1.69 (1.18-2.43) | 0.026, 1.56 (1.06-2.29) | 0.005, 1.95 (1.22-3.12) | 0.03, 1.73 (1.06-2.83) |
| ≥60 | <0.001, 2.55 (1.63-4.01) | 0.003, 2.07 (1.27-3.38) | <0.001, 4.08 (1.92-8.66) | 0.01, 2.87 (1.28-6.40) | <0.001, 3.94 (2.76-5.61) | <0.001, 3.09 (2.12-4.52) | <0.001, 4.39 (2.79-6.89) | <0.001, 3.08 (1.90-5.00) |
| Race | ||||||||
| white | reference | reference | reference | reference | ||||
| black | 0.712, 1.10 (0.67-1.80) | 0.702, 0.94 (0.68-1.30) | 0.840, 1.10 (0.44-2.78) | 0.491, 1.39 (0.54-3.56) | 0.427, 0.84 (0.55-1.29) | 0.646, 0.90 (0.56-1.43) | 0.289, 0.74 (0.42-1.30) | 0.396, 0.76 (0.40-1.44) |
| other | 0.95, 0.99 (0.74-1.33) | 0.974, 1.01 (0.55-1.87) | 0.800, 1.13 (0.45-2.85) | 0.796, 0.84 (0.36-2.73) | 0.106, 1.25 (0.95-1.63) | 0.081, 1.30 (0.97-1.75) | 0.16, 1.27 (0.91-1.76) | 0.100, 1.36 (0.94-1.97) |
| Sex | ||||||||
| female | reference | reference | reference | reference | ||||
| male | 0.01, 1.36 (1.08-1.71) | 0.015, 1.39 (1.07-1.80) | 0.347, 1.31 (0.75-2.29) | 0.243, 1.45 (0.78-2.69) | <0.001, 1.73 (1.41-2.11) | <0.001, 1.80 (1.44-2.26) | <0.001, 1.75 (1.35-2.26) | <0.001, 1.71 (1.28-2.30) |
| Grade | ||||||||
| low-grade | reference | reference | reference | reference | ||||
| high-grade | 0.145, 1.52 (0.87-2.67) | 0.499, 1.22 (0.69-2.16) | 0.058, 3.88 (0.96-15.77) | 0.0307, 7.13 (1.12-45.23) | 0.001, 2.16 (1.34-3.46) | 0.005, 2.11 (1.25-3.55) | 0.008, 2.32 (1.25-4.32) | 0.026, 2.24 (1.10-4.54) |
| AFP | ||||||||
| normal | reference | reference | reference | reference | ||||
| elevated | 0.801, 1.11 (0.51-2.40) | 0.878, 0.93 (0.38-2.31) | 0.119, 2.76 (0.77-9.90) | 0.257, 2.43 (0.52-11.23) | 0.686, 1.14 (0.60-2.20) | 0.878, 0.94 (0.44-2.04) | 0.462, 1.37 (0.59-3.19) | 0.792, 1.14 (0.41-3.20) |
| Fibrosis | ||||||||
| none to moderate | reference | reference | reference | reference | ||||
| Severe | 0.067, 1.53 (0.62-3.78) | 0.058, 2.829 (0.80-1.06) | 0.022, 10.07 (1.39-73.07) | 0.014, 12.35 (1.68-90.82) | 0.017, 2.60 (1.19-5.70) | 0.006, 4.36 (1.53-11.87) | 0.123, 2.09 (0.82-5.31) | 0.096, 2.85 (0.83-9.78) |
| Tumor number | ||||||||
| 1 | reference | reference | reference | reference | ||||
| ≥2 | 0.684, 0.93 (0.66-1.32) | <0.001, 0.041 (0.01-0.22) | 0.09, 2.06 (0.97-4.38) | 0.215, 0.044 (0.00-6.18) | 0.545, 1.11 (0.80-1.53) | <0.001, 0.04 (0.01-0.22) | 0.399, 1.19 (0.80-1.78) | 0.004, 0.04 (0.01-0.36) |
| Tumor size | ||||||||
| ≤5 cm | reference | reference | reference | reference | ||||
| >5 cm | 0.03, 1.70 (1.05-2.75) | 0.075, 1.67 (0.95-2.96) | 0.003, 4.79 (1.73-13.25) | 0.004, 6.01 (1.80-20.07) | <0.001, 2.86 (1.90-4.32) | <0.001, 2.81 (1.74-4.54) | <0.001, 2.90 (1.77-4.74) | 0.003, 2.39 (1.35-4.25) |
| Surgery | ||||||||
| Yes | reference | reference | reference | reference | ||||
| No | 0.016, 2.20 (1.16-4.18) | 0.062, 1.96 (0.97-3.97) | 0.814, 1.11 (0.47-2.63) | 0.803, 0.89 (0.37-2.16) | 0.521, 1.20 (0.69-2.07) | 0.488, 1.24 (0.67-2.30) | 0.854, 1.09 (0.47-2.47) | 0.893, 1.07 (0.41-2.78) |
| Radiation | ||||||||
| Yes | reference | reference | reference | reference | ||||
| No | 0.411, 1.28 (0.71-2.35) | 0.35, 1.40 (0.69-2.83) | 0.164, 0.37 (0.09-1.51) | 0.588, 0.58 (0.08-4.22) | 0.625, 0.87 (0.50-1.51) | 0.998, 0.99 (0.52-1.94) | 0.762, 1.12 (0.53-2.38) | 0.412, 1.51 (0.56-4.08) |
| Chemotherapy | ||||||||
| Yes | reference | reference | reference | reference | ||||
| 0.033, 1.41 (1.03-1.93) | 0.117, 1.32 (0.93-1.87) | 0.996, 1.00 (0.49-2.05) | 0.476, 0.77 (0.37-1.60) | 0.208, 1.19 (0.91-1.55) | 0.417, 1.13 (0.84-1.52) | 0.267, 1.20 (0.87-1.66) | 0.533, 1.12 (0.78-1.62) | |
| Histological type | ||||||||
| AS | reference | reference | ||||||
| HE | 0.019, 2.33 (1.15-4.72) | 0.034, 2.41 (1.07-5.45) | 0.037, 2.90 (1.06-7.90) | 0.074, 2.88 (0.90-9.18) | ||||
| EHE | 0.896, 0.95 (0.43-2.09) | 0.890, 0.94 (0.38-2.34) | 0.783, 1.17 (0.39-3.50) | 0.921, 1.07 (0.29-3.83) | ||||
| HP | 0.077, 0.51 (0.24-1.08) | 0.160, 0.54 (0.23-1.27) | 0.258, 0.55 (0.19-1.56) | 0.361, 0.57 (0.17-1.92) | ||||
| T stage | ||||||||
| early stage | reference | reference | reference | reference | ||||
| advanced stage | 0.053, 1.45 (1.00-2.11) | 0.113, 1.43 (0.92-2.22) | 0.025, 2.60 (1.13-5.98) | 0.016, 3.03 (1.23-7.46) | 0.001, 1.74 (1.25-2.42) | 0.003, 1.80 (1.23-2.63) | 0.008, 1.75 (1.16-2.63) | 0.024, 1.76 (1.08-2.88) |
| N stage | ||||||||
| N0 | reference | reference | reference | reference | ||||
| N1 | 0.206, 1.59 (0.78-3.24) | 0.135, 1.80 (0.83-3.87) | 0.142, 1.96 (0.80-4.84) | 0.036, 2.72 (1.07-6.92) | 0.711, 1.11 (0.65-1.88) | 0.454, 1.24 (0.70-2.20) | 0.662, 0.85 (0.41-1.75) | 0.683, 0.841 (0.37-1.93) |
| M stage | ||||||||
| M0 | reference | reference | reference | reference | ||||
| M1 | 0.089, 1.33 (0.96-1.83) | 0.124, 1.34 (0.92-1.93) | 0.080, 1.90 (0.93-3.87) | 0.073, 2.08 (0.93-4.64) | 0.198, 1.16 (0.88-1.54) | 0.277, 1.19 (0.87-1.64) | 0.513, 1.13 (0.79-1.62) | 0.419, 1.19 (0.79-1.79) |
HR: Hazard Ratio. CI: Confidence Interval. Low-grade included grades I and II. Grade III and IV were classified into high-grade. T1 and T2 were included in the early stage. T3 and T4 were included in the advanced stage. AS: angiosarcoma. HE: hemangioendothelioma. EHE: epithelioid hemangioendothelioma. HP: hemangiopericytoma.
Next, the KM method was performed to calculate the survival probability of OS (Figure 2) and CSS (Figure 3) for patients in the AS, EHE, and HMVT-all cohorts classified by the variables showing a significant correlation with prognosis. Correspondingly, we analyzed the median OS and CSS associated with these variables in the three cohorts, as shown in Table S1.
Figure 2.
The Kaplan-Meier (KM) method was performed to calculate the OS classified by clinical pathologic factors for patients from diverse HMVT cohorts including AS (A-G), EHE (H-K), and HMVT-all (L-Q).
Figure 3.
The Kaplan-Meier (KM) method was performed to calculate the CSS classified by clinical pathologic factors for patients within diverse HMVT cohorts including AS (A-D), EHE (E-I), and HMVT-all (J-N).
Multivariate analysis of the independent prognostic factors for OS and CSS
To adjust for the interaction between various covariates, the multivariate Cox proportional hazard model of AS, EHE, HMVT-all, and -training cohort was conducted to determine the independent risk factors associated with OS and CSS (Table 3). The factors with a P-value of <0.05 and several significant clinicopathological indicators, such as surgery, chemotherapy, fibrosis, and TNM stages, were all included in the Cox multivariate analysis. For AS patients, results indicated that age ≥60 years (HR=2.46, 95% CI: 1.54-3.92, P<0.001), sex (HR=1.24, 95% CI: 0.97-1.57, P=0.043), surgery (HR=2.16, 95% CI: 1.12-4.17, P=0.022), and chemotherapy (HR=1.48, 95% CI: 1.06-2.06, P=0.021) were independent predictors of OS. Simultaneously, they are also the prognostic factors for predicting inferior CSS in patients within AS cohort. Moreover, the advanced T stage also exhibited an inferior influence on OS (HR=0.54, 95% CI: 0.28-1.02, P=0.059).
Table 3.
Multivariate analysis of OS and CSS for patients with hepatic vascular tumor (HMVT)
| Categories | AS (N=350) | EHE (N=114) | HMVT-all (N=510) | HMVT-training (N=308) | ||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| P-value, HR (95% CI) | ||||||||
|
| ||||||||
| OS | CSS | OS | CSS | OS | CSS | OS | CSS | |
| Age | ||||||||
| <40 | reference | reference | reference | reference | ||||
| 40-59 | 0.240, 1.33 (0.83-2.16) | 0.554, 1.17 (0.70-1.96) | 0.548, 1.28 (0.57-2.91) | 0.965, 0.98 (0.40-2.38) | 0.119. 1.34 (0.93-1.94) | 0.287, 1.24 (0.84-1.84) | 0.063, 1.57 (0.98-2.59) | 0.224, 1.37 (0.82-2.28) |
| ≥60 | <0.001, 2.46 (1.54-3.92) | 0.014, 1.87 (1.13-3.08) | <0.001, 5.41 (2.41-12.16) | 0.003, 3.91 (1.59-9.63) | <0.001, 3.09 (2.14-4.46) | <0.001, 2.49 (1.68-3.71) | <0.001, 3.27 (2.03-5.19) | 0.002, 2.27 (1.36-3.78) |
| Sex | ||||||||
| female | reference | reference | reference | reference | ||||
| male | 0.043, 1.24 (0.97-1.57) | 0.042, 1.33 (1.01-1.74) | 0.116, 1.18 (0.96-1.46) | 0.030, 1.30 (1.03-1.65) | 0.155, 1.22 (0.93-1.60) | 0.119, 1.28 (0.94-1.75) | ||
| Grade | ||||||||
| low-grade | reference | reference | reference | reference | ||||
| high-grade | 0.072, 6.62 (0.85-51.86) | 0.023, 1.78 (1.08-2.91) | 0.042, 1.77 (1.02-3.08) | 0.089, 1.75 (0.92-3.32) | 0.137, 1.75 (0.84-3.68) | |||
| Fibrosis | ||||||||
| none to moderate | reference | reference | reference | reference | ||||
| severe | 0.311, 1.64 (0.63-4.24) | 0.973, 0.98 (0.27-3.55) | 0.009, 17.92 (2.04-157.43) | 0.009, 30.33 (2.38-386.40) | 0.334, 1.49 (0.66-3.34) | 0.567, 1.37 (0.47-3.97) | 0.363, 0.55 (0.15-2.01) | |
| Tumor number | ||||||||
| 1 | reference | reference | reference | reference | ||||
| ≥2 | 0.921, 0.00 (0.00-infinite) | 0.917, 0.00 (0.00-0.005) | 0.935, 0.00 (0.00-infinite) | |||||
| Tumor size | ||||||||
| ≤5 cm | reference | reference | reference | reference | ||||
| >5 cm | 0.595, 1.18 (0.64-2.16) | 0.964, 0.98 (0.45-2.15) | 0.016, 0.24 (1.30-12.91) | 0.009, 6.47 (1.61-26.06) | 0.039, 1.55 (0.97-2.50) | 0.003, 2.38 (133-4.26) | 0.009, 2.21 (1.22-4.000) | 0.043, 2.21 (1.02-4.44) |
| Surgery | ||||||||
| Yes | reference | reference | reference | reference | ||||
| No | 0.022, 2.16 (1.12-4.17) | 0.047, 2.08 (1.01-4.27) | ||||||
| Chemotherapy | ||||||||
| Yes | reference | reference | reference | reference | ||||
| No | 0.021, 1.48 (1.06-2.06) | 0.023, 1.54 (1.06-2.22) | ||||||
| Histological type | ||||||||
| AS | reference | reference | ||||||
| HE | <0.001, 0.47 (0.31-0.71) | 0.001, 0.46 (0.28-0.74) | 0.004, 0.45 (0.26-0.77) | 0.001, 0.34 (0.18-0.66) | ||||
| EHE | <0.001, 0.30 (0.22-0.41) | <0.001, 0.30 (0.21-0.43) | <0.001, 0.27 (0.18-0.41) | <0.001, 0.23 (0.14-0.38) | ||||
| HP | 0.131, 0.58 (0.28-1.18) | 0.314, 0.65 (0.28-1.50) | 0.086, 0.41 (0.15-1.14) | 0.185, 0.45 (0.14-1.47) | ||||
| T stage | ||||||||
| early stage | reference | reference | reference | reference | ||||
| advanced stage | 0.059, 0.54 (0.28-1.02) | 0.044, 2.77 (1.03-7.44) | 0.201, 2.19 (0.66-7.32) | 0.041, 1.50 (1.02-2.20) | 0.128, 1.42 (0.90-2.22) | 0.464, 1.21 (0.73-1.99) | 0.891, 1.05 (0.55-1.97) | |
| N stage | ||||||||
| N0 | reference | reference | reference | reference | ||||
| N1 | 0.232, 2.48 (0.56-11.02) | |||||||
| M stage | ||||||||
| M0 | reference | reference | reference | reference | ||||
| M1 | 0.110, 1.31 (0.94-1.84) | 0.072, 1.43 (0.97-2.10) | 0.028, 2.34 (1.10-4.98) | 0.039, 2.81 (1.06-7.47) | ||||
Among patients with EHE, age ≥60 years (HR=5.41 and P<0.001; HR=3.91 and P=0.003), severe fibrosis (HR=17.92 and P=0.009; HR=30.33 and P=0.009), tumor size >5 cm (HR=0.24, P=0.016; HR=6.47, P=0.009), and M1 stage (HR=2.34 and P=0.028; HR=2.81 and P=0.039) were associated with poor OS and CSS. Furthermore, the advanced T stage (HR=2.77, 95% CI: 1.03-7.44, P=0.044) was a significant risk factor of OS in patients with EHE. In the HMVT-all cohort, age ≥60 years (HR=3.09 and P<0.001; HR=2.49 and P<0.001), high tumor grade (HR=1.78 and P=0.023; HR=1.77 and P=0.042), and larger tumor size (HR=1.55, P=0.039; HR=2.38, P=0.003) were negatively correlated with both OS and CSS. Similarly, the histological type is another significant independent predictor of survival time for patients with HMVT. In addition, the prognostic difference was found in the OS between the patients diagnosed at advanced T stage and those at early T stage in the HMVT-all cohort, with a P-value of 0.041. Concurrently, the risk factors including age, tumor size, and histological type were also the independent predictors for OS and CSS in patients within HMVT-training group.
Establishment and evaluation of the prognostic nomograms
To evaluate survival probability for individual patients, the predictive nomogram models for OS and CSS in the HMVT-training cohort were established. The significantly prognostic factors screened from the multivariate analysis including age, sex, T stage, tumor size, and histological type were integrally operated. However, the risk factor of the grade was excluded since the number of patients with a clear grade ranking was not enough for model construction (Figure 4A and 4B). The detailed points of each factor were presented in the nomogram, the 1-, 3- and 5-year OS and CSS could be predicted by calculating these scores to the total on the bottom axis. In the training cohort, the C-index for OS and CSS was 0.763 (95% CI: 0.708-0.819) and 0.722 (95% CI: 0.653-0.791) respectively. Meanwhile, the calibration curves estimating the 1-, 3- and 5-year OS and CSS rates showed an optimal agreement between the nomogram-predicted and the actual observation (Figure 4C and 4D). Moreover, the ROC was performed to assess the accuracy of the prognostic models. The AUC value for 1-, 3- and 5-year OS in the training cohort was 0.873, 0.905, and 0.898. The corresponding AUC value for CSS was 0.808, 0.794, and 0.788 (Figure 4E and 4F). These data indicated that the nomograms based on HMVT-training cohort exhibited an authentically predictive ability.
Figure 4.
The establishment and evaluation of clinical prognostic nomogram models to predict 1-year, 3-year, and 5-year of OS and CSS for patients via HMVT-training cohort. (A and B) The nomogram for OS and CSS respectively. The calibration and area under the receiver operating characteristic curve (AUC) of 1-year, 3-year, and 5-year to assess the performance of predictive models for OS (C and E) and CSS (D and F). The function of this nomogram, the value of each variable for an individual patient is recorded on the corresponding axis, and the number of points obtained for each variable value is determined by drawing a line upward to the point axis. The overall scores for each patient could be found in the corresponding location in the axis of “Total Points axis”. Then a line is drawn downward to the survival axes to determine the possibility of a 1-, 3- or 5-year survival rate. ROC: receiver operating characteristic curve. TP: true positive. FP: false positive.
Subsequently, the validation for the nomograms was executed in various test cohorts to further verify the reliability. The C-index values for predicting OS and CSS were 0.741 (95% CI: 0.661-0.745) and 0.743 (95% CI: 0.653-0.833) in HMVT-validation, 0.751 (95% CI: 0.704-0.798) and 0.731 (95% CI: 0.677-0.786) in HMVT-all (Table 4). The analysis of AUC and the calibration curves in these validated cohorts showed the consistent results that have been observed in the training cohort (Table 4; Figure 5). Collectively, these results confirmed that the established nomograms were strongly convinced and valuable in clinical practice to predict prognosis for HMVT patients. Furthermore, the nomograms for OS in the histologic type of AS and EHE were developed and the accuracy was assessed as well, exhibiting credible discrimination (Figure S3).
Table 4.
The detailed information of C-index and AUC for patients with HMVT within the HMVT-training, validation, and all cohort
| Survival types | cohort | C-index | AUC | ||
|---|---|---|---|---|---|
|
|
|||||
| (value, 95% CI) | 1-year | 3-year | 5-year | ||
| OS | HMVT-training | 0.763 (0.708-0.819) | 0.873 | 0.905 | 0.898 |
| HMVT-validation | 0.741 (0.661-0.745) | 0.788 | 0.843 | 0.843 | |
| HMVT-all | 0.751 (0.704-0.798) | 0.826 | 0.89 | 0.886 | |
| CSS | HMVT-training | 0.722 (0.653-0.791) | 0.808 | 0.794 | 0.788 |
| HMVT-validation | 0.743 (0.653-0.833) | 0.762 | 0.909 | 0.909 | |
| HMVT-all | 0.731 (0.677-0.786) | 0.781 | 0.895 | 0.885 | |
C-index: concordance index. AUC: area under the receiver operating character curve. CI: confidence interval. OS: overall survival. CSS: cancer-specific survival.
Figure 5.
The calibration plots for 1-year, 3-year, and 5-year of OS (A) and CSS (B) respectively in patients from HMVT-validation and HMVT-all cohort.
Next, we quantitatively calculated the risk score based on survival-related clinicopathologic factors (age, sex, grade, tumor size, and T stage) in HMVT pa-tients. To further ex-plore whether risk score could be regarded as an independent prognostic trait, we determined the median risk scores as the cutoff value to divide the patients in the HMVT-training cohorts into low-risk and high-risk groups. The OS and CSS for patients at different risk groups were analyzed using KM curves classified by clinicopathologic factors and histological types. For the analysis of OS, the patients included in the high-risk group had a shorter survival time than those who are in the low-risk group in most variables except for in the stratification of HE histologic type (Figure 6). Similarly, when the CSS was evaluated stratified by age at 40-59 years, sex, grade, T stage, and AS type, a longer survival time was observed in the patients within the low-risk group instead of those in the high-risk group (Figure 7).
Figure 6.
The overall survival (OS) in patients within the HMVT-training cohort in different risk stratification classified by clinicopathologic factors and tumor types. A: Risk score. B: Histological type. C: Age. D: Sex. E: Grade. F: Tumor size. G: T stage.
Figure 7.
The cancer-specific survival (CSS) in patients within the HMVT-training cohort in different risk stratification classified by clinicopathologic factors and tumor types. A: Risk score. B: Histological type. C: Age. D: Sex. E: Grade. F: T stage. G: Tumor size. The patients with tumor size of <5 cm are all included in the low-risk group.
Discussion
Due to the rarity of HMVT, only a few studies investigated the profile of demographic traits, clinical manifestation, and survival probability of these patients [17-19]. SEER database, maintained by the US National Cancer Institute, is the largest institution that collects and reports cancer survival and incidence data. Since this database has the feature of high quality and collecting clinical cases in a uniformly standard criterion, it would give rise to a low rate of errors and make these recorded documents more accurate and reliable. To better understand the clinicopathological traits and risk factors of this disease, we investigated the largest sample of patients with HMVT retrieved from the SEER database.
In this study, a total of 510 patients diagnosed with HMVT between 1973 and 2015 were extracted from the SEER database, which included those with AS (N=350), HE (N=37), EHE (N=114), and HP (N=9). Results revealed that the 1-, 3- and 5-year OS and CSS in the overall HMVT cohort were 28.9% and 37.9%, 21.3% and 29.8%, and 19.8% and 27.7% respectively. Concerning the specific types of HMVT, AS had the lowest 3- and 5-year OS and CSS compared with HP, EHE, and HE. HP cases were sporadic and only 9 patients were reported, of whom 5 survived 5 years after diagnosis. Recently, a retrospective study, including 42 patients with pathologically confirmed HMVT, was conducted. It was indicated that 13 patients with HE and 15 with EHE had the longest survival time, with a follow-up time of 96 and 88 months respectively, followed by 3 patients with HP (23 months). By contrast, 11 patients with AS experienced the shortest survival time (15 months) [19].
In our study, we found that the clinicopathologic traits of age ≥60 years, high tumor grade, larger tumor size (>5 cm), advanced T stage, and histological type are independent prognostic predictors associated with poorer OS in the HMVT-all cohort. However, the M stage is not a predictor of prognosis for patients in the HMVT-all cohort. In separate analyses, we further noticed that the M stage is an independent indicator of predicting survival probability for individual patients with EHE, although it was not significantly associated with prognosis in the AS cohort. Then, we reviewed the currently published researches relevant to AS and found that patients are easily diagnosed at a late stage with larger lesions of >5 cm since many patients didn’t have any notably obvious anomalies in the early stage. Meanwhile, secondary symptoms including tumor bleeding, tumor rupture, and other complications (such as severe pneumonia, multisystem organ failure, and recurrence after surgery) are preferable to occur in AS patients at late stages. These factors could result in patients’ death no matter whether or not distant metastasis is presented [20-23]. Zeng et al. performed a systematic review of the published studies worldwide from 1990 to 2019 to comprehensively analyze the main clinical traits, demographics, therapeutic strategy, and prognosis of 219 patients with AS. They also indicated that although patients with metastasis at the time of diagnosis had a significantly shorter survival time, the metastasis didn’t show independent prognostic value in multivariate Cox regression. The patients who have tumor rupture would have an obviously worse prognosis [24].
Subsequently, the prognostic nomograms were established for estimating OS and CSS exhibiting a credible performance with the C-index of 0.763 and 0.722, respectively. The predictive models could be used to calculate the 1-, 3- and 5-year OS for individual patients with HMVT, with great importance in clinical practice. According to the findings in our study, we could know that the survival probability in patients with HMVT is different from hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA) that ranked the first and second most common liver cancer, respectively. For the patients diagnosed with HCC or CCA at an early stage, the curative treatments including surgical resection, liver transplantation, and ablation were widely considered to be the first-line strategies. It was reported that the 5-year OS for HCC patients referred to 70%, and the median survival time underlying intention-to-treat analysis of CCA is 36 months. However, the recurrence rate after surgery at 5-years in HCC patients was up to 70%, correspondingly, 65% for CAA. Furthermore, more than 70% of patients with HCC and CAA were at advanced stages when diagnosed, with only being left chance of palliative treatments such as transcatheter arterial chemoembolization (TACE), chemotherapy, and molecular targeted therapeutic strategies [25,26].
Among various histological types of HMVT, AS showed the highest incidence, accounting for one-third of primary sarcomas. Histologically, AS is composed of endothelial cells with variable differentiation ability [27,28]. Several chemical factors such as vinyl chloride, arsenic, thorotrast, and thorium dioxide were intimately associated with the development of AS, but most cases are still idiopathic and are reported worldwide [28,29]. Molina E et al. described the survival probability of AS in five case reports, of which two patients underwent surgical treatment, and another two patients were found to have unresectable bilateral disease intraoperatively. The median survival time of the five patients was 6 months (range: 3 days to 18 months) [27]. Another retrospective study, including 60 patients with AS, indicated that the OS rate at 5 years was 20%, and surgical therapy was a significant predictor for better prognosis [30]. Although these studies provided us with some meaningful information on the survival probability of AS, due to the small sample size, knowledge on its clinicopathological traits was still limited and the prognostic predictors associated with the clinical outcomes were not explored yet. Therefore, a large sample of AS patients was used in the current study by acquiring reliable demographic and clinicopathological data from the SEER database. In this population-based study, we found that most cases with AS were diagnosed in patients aged 55-74 years, and AS infrequently occurred in those aged ≤34 years, which was consistent with the previous reports [27,30,31]. Meanwhile, the 1-, 3-, and 5-year OS rates were 12.6%, 5.9%, and 5.9%, respectively. The potential predictors associated with OS for AS cases were including age ≥60 years, male sex, surgery, and chemotherapy. Whereas, neither the treatment of surgery and chemotherapy nor radiation indicated the independent prognostic value in the HMVT-all cohort. The low ratio of the patients with the clear medical document of therapeutic strategies may be the potential limitation leading to the reduced analysis performance of multivariate Cox regression. In the further, the large population of patients with HMVT including treatment information is still requested to precisely elaborate the role of treatments in predicting survival probability.
HE is an extremely rare type of tumor, and its incidence rate is lower than that of AS among HMVT, the majority of previous studies on HE indicated the epithelioid type (EHE). It originates from the soft tissues and rarely occurs in the liver, suggesting its intermediate and malignant potential, although the World Health Association classification still considers EHE as a malignant sarcoma [32,33]. The morphological features of EHE were first described by Weiss and Enzinger in 1982. The main components of EHE included rounded or slightly spindled eosinophilic endothelial cells and obvious cytoplasmic vacuolization. The absence of pleomorphism and mitotic activity in most cases and the focal features of vascular channels could distinguish this neoplasm from a carcinoma, although its growth pattern is similar to that of solid tumors and exhibits an epithelioid appearance in the endothelium [34]. The first series of hepatic EHE cases (32 cases) were reported by Ishak et al. in 1984, which described the clinical, morphologic, and follow-up data [35]. Based on this retrospective study, women (62.5%) were more affected than men (37.5%), and the average age was 49.65 years (range: 19-86 years). Meanwhile, nine patients were followed-up for 5 years or longer after the initial diagnosis. Another single institution from Pittsburgh also studied a small cohort of hepatic EHE patients, including 25 cases who either underwent surgery or embolization. The recorded median OS for all patients was 13.9 years. Two patients who underwent partial resection survived after a follow-up of 19 and 71 months respectively. Among 17 patients who had a liver transplant, lymphatic metastasis and vascular invasion were found in 6 patients (35.3%) and 11 patients (64.7%), respectively [36].
Although EHE is a rare malignant sarcoma, the number of EHE cases that occurred in various organs including the liver, are gradually increasing with the robust improvement of diagnostic techniques. To some extent, these sporadic reports contribute to our understanding of this neoplasm, such as symptoms, preferable age at diagnosis, survival probability, effective treatments, and predictors of OS and CSS. However, these studies usually were conducted in small sample series or were case reports, and some information was contradictory since the methods of diagnosis or treatment varied in various hospitals. To further systematically describe the clinicopathological characteristics of hepatic EHE, a large sample cohort was required in future retrospective analyses. Therefore, we collected the clinical documents of patients diagnosed with hepatic EHE who were included in the SEER database between 1973 and 2015, all cases were registered according to a uniform standard. The 3- and 5-year OS rates of hepatic EHE were 60.5% and 54.8%, respectively, showing better survival trends than AS. This finding was also consistent with those of previous studies [32,37,38]. Moreover, we found that age ≥60 years, fibrosis level, tumor size, T stage, and M stage were risk factors associated with patients’ prognosis. The prognostic nomogram model also demonstrated its reliable predictive performance that could help clinicians assess the progression of the disease and evaluate the survival probability at 3-, and 5-year for individual patients.
HP is the rarest type of HMVT derived from pericytes, which are wrapped by blood vessels and capillaries. HP commonly develops in the lower extremities, pelvis, meninges, and lung, but occurs less likely in the breast, bones, liver, and stomach [5,39,40]. Previous studies revealed that this tumor type is more common in middle-aged patients than in infants and children. Generally, a biopsy or tumor resection is required to verify the histopathological type, however, it is usually misdiagnosed as another type of soft tissue sarcoma due to the lack of classical traits. Therefore, the diagnosis of HP in clinical practice often relies on exclusion criteria [5,41]. In the current study, we finally extracted the clinical data of 9 patients diagnosed with hepatic HP, the 3-and 5-year OS rates were 33.3% (3 patients) and 33.3% (3 patients). Among nine patients, only one was diagnosed at the age of <40 years, and six were diagnosed at the age of 40-59 years. The findings were similar to those reported in previous studies, indicating that HP commonly affected older adults [40,41]. To date, surgery remains the recommended treatment for HP patients. As expected, the OS and CSS in patients with HP could be improved by surgery. However, disease recurrence after surgery remains a challenge, with an incidence of more than 30%. In this cohort, only one patient with HP underwent surgical resection and survived for 8 months.
In this study, all clinical data were carefully extracted from the same database. Uniform standards and approaches were used to evaluate practical conditions for individual patients from various aspects, which supplied strong statistical power for this retrospective study and allowed us to comprehensively analyze the clinicopathological traits of rare tumors such as HMVT. However, there are still some limitations. First, the ratio of the number of patients in the HMVT cohort with specific treatment information is low, which restricted the further precise analyses. Second, since the SEER database contained qualitative or semiquantitative data, the statistical reliability was compromised to some extent. Finally, biases were inevitably introduced due to the nature of the retrospective research.
In all, this is the largest population-based retrospective study to describe the demographic and clinicopathological characteristics of HMVT, especially concerning AS, HE, EHE, and HP. Univariate and multivariate analyses were performed to identify independent predictors associated with OS and CSS including age ≥60 years, high tumor grade, tumor size (>5 cm), and the histological type. Besides, the advanced T stage was the prognostic factor for prediction of OS in the HMVT cohort. The prognostic nomograms for estimating 1-, 3- and 5-year OS and CSS indicated a credible performance. Although our findings and novel model need to be improved by conducting further in-depth research, our results may provide new insight regarding the features and outcomes of patients with HMVT.
Acknowledgments
We put much appreciation to all the staff from the Surveillance, Epidemiology, and End Results (SEER) database that provide us the chance to analyze these valuable clinical documents of patients. This work was supported by grants from the National Natural Sciences Foundation of China (No.81970523, No. 81800506), the National Science and Technology Major Project of China (2018ZX10732202, 2017YFC0908104), Natural Sciences Foundation of Hunan province (No. 2020JJ4877, No. 2020JJ5886, No. 2020JJ4906, No. 2019JJ40496), Basic and Applied Basic Research Foundation of Guangdong Province (No. 2020A1515110491).
Disclosure of conflict of interest
None.
Abbreviations
- HMVT
Hepatic malignant vascular tumors
- OS
overall survival
- CSS
cancer-specific survival
- AS
angiosarcoma
- HE
hemangioendothelioma
- EHE
epithelioid hemangioendothelioma
- HP
hemangiopericytoma
- MVT
Malignant vascular tumors
- SEER database
Surveillance, Epidemiology, and End Results database
- AFP
alpha-fetoprotein protein
- TNM
tumor-node-metastasis
- AUC
area under the curve
- ROC
receiver operating characteristic
- HRs
Hazard ratios
- CIs
confidence intervals
Supporting Information
References
- 1.Groeschl RT, Miura JT, Oshima K, Gamblin TC, Turaga KK. Does histology predict outcome for malignant vascular tumors of the liver? J Surg Oncol. 2014;109:483–486. doi: 10.1002/jso.23517. [DOI] [PubMed] [Google Scholar]
- 2.Hu S, Wang X. Prognostic determinants analysis and nomogram for bone malignant vascular tumors: a surveillance, epidemiology and end results (SEER) analysis. Med Sci Monit. 2020;26:e923305. doi: 10.12659/MSM.923305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wang W, Hong J, Meng J, Wu H, Shi M, Yan S, Huang Y. Survival analysis of patients with osseous malignant vascular tumors: results of the surveillance, epidemiology, and end results (SEER) database from 1973 to 2015. Med Sci Monit. 2019;25:5525–5535. doi: 10.12659/MSM.914950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Malmgren JA, Calip GS, Atwood MK, Mayer M, Kaplan HG. Metastatic breast cancer survival improvement restricted by regional disparity: surveillance, epidemiology, and end results and institutional analysis: 1990 to 2011. Cancer. 2020;126:390–399. doi: 10.1002/cncr.32531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wang K, Mei F, Wu S, Tan Z. Hemangiopericytoma: incidence, treatment, and prognosis analysis based on SEER database. Biomed Res Int. 2020;2020:2468320. doi: 10.1155/2020/2468320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Young RJ, Brown NJ, Reed MW, Hughes D, Woll PJ. Angiosarcoma. Lancet Oncol. 2010;11:983–991. doi: 10.1016/S1470-2045(10)70023-1. [DOI] [PubMed] [Google Scholar]
- 7.Wenger DE, Wold LE. Malignant vascular lesions of bone: radiologic and pathologic features. Skeletal Radiol. 2000;29:619–631. doi: 10.1007/s002560000261. [DOI] [PubMed] [Google Scholar]
- 8.Cioffi A, Reichert S, Antonescu CR, Maki RG. Angiosarcomas and other sarcomas of endothelial origin. Hematol Oncol Clin North Am. 2013;27:975–988. doi: 10.1016/j.hoc.2013.07.005. [DOI] [PubMed] [Google Scholar]
- 9.Stout AP. Hemangiopericytoma; a study of 25 cases. Cancer. 1949;2:1027–1054. doi: 10.1002/1097-0142(194911)2:6<1027::aid-cncr2820020609>3.0.co;2-r. illust. [DOI] [PubMed] [Google Scholar]
- 10.Shaigany K, Fang CH, Patel TD, Park RC, Baredes S, Eloy JA. A population-based analysis of head and neck hemangiopericytoma. Laryngoscope. 2016;126:643–650. doi: 10.1002/lary.25681. [DOI] [PubMed] [Google Scholar]
- 11.Kamamoto D, Ohara K, Kitamura Y, Yoshida K, Kawakami Y, Sasaki H. Association between programmed cell death ligand-1 expression and extracranial metastasis in intracranial solitary fibrous tumor/hemangiopericytoma. J Neurooncol. 2018;139:251–259. doi: 10.1007/s11060-018-2876-7. [DOI] [PubMed] [Google Scholar]
- 12.Kulshreshtha P, Kannan N, Bhardwaj R, Batra S, Gupta S. Primary mediastinal hemangiopericytoma treated with preoperative embolization and surgery. Ann Thorac Surg. 2014;97:335–338. doi: 10.1016/j.athoracsur.2013.04.128. [DOI] [PubMed] [Google Scholar]
- 13.Bokshan SL, Doyle M, Becker N, Nalbantoglu I, Chapman WC. Hepatic hemangiopericytoma/solitary fibrous tumor: a review of our current understanding and case study. J Gastrointest Surg. 2012;16:2170–2176. doi: 10.1007/s11605-012-1947-x. [DOI] [PubMed] [Google Scholar]
- 14.Caruso S, Gruttadauria S, Minervini MI, Miraglia R, Milazzo M, Luca A, Gridelli B. Primary hemangiopericytoma of the liver: sonographic findings. J Clin Ultrasound. 2009;37:305–307. doi: 10.1002/jcu.20567. [DOI] [PubMed] [Google Scholar]
- 15.Koch M, Nielsen GP, Yoon SS. Malignant tumors of blood vessels: angiosarcomas, hemangioendotheliomas, and hemangioperictyomas. J Surg Oncol. 2008;97:321–329. doi: 10.1002/jso.20973. [DOI] [PubMed] [Google Scholar]
- 16.Apra C, Mokhtari K, Cornu P, Peyre M, Kalamarides M. Intracranial solitary fibrous tumors/hemangiopericytomas: first report of malignant progression. J Neurosurg. 2018;128:1719–1724. doi: 10.3171/2017.1.JNS162593. [DOI] [PubMed] [Google Scholar]
- 17.Bioulac-Sage P, Laumonier H, Laurent C, Blanc JF, Balabaud C. Benign and malignant vascular tumors of the liver in adults. Semin Liver Dis. 2008;28:302–314. doi: 10.1055/s-0028-1085098. [DOI] [PubMed] [Google Scholar]
- 18.Zhang P, Hu J, Zhou D. Hemangiopericytoma of the cervicothoracic spine: a case report and literature review. Turk Neurosurg. 2014;24:948–953. doi: 10.5137/1019-5149.JTN.8299-13.1. [DOI] [PubMed] [Google Scholar]
- 19.Zhou Y, Hou P, Wang F, Li B, Gao J. Primary hepatic malignant vascular tumors: a follow-up study of imaging characteristics and clinicopathological features. Cancer Imaging. 2020;20:59. doi: 10.1186/s40644-020-00336-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gigante E, Paradis V, Ronot M, Cauchy F, Soubrane O, Ganne-Carrie N, Nault JC. New insights into the pathophysiology and clinical care of rare primary liver cancers. JHEP Rep. 2021;3:100174. doi: 10.1016/j.jhepr.2020.100174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Jiang L, Xie L, Li G, Xie H, Fang Z, Cai X, Chen Y. Clinical characteristics and surgical treatments of primary hepatic angiosarcoma. BMC Gastroenterol. 2021;21:156. doi: 10.1186/s12876-021-01743-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Molina E, Hernandez A. Clinical manifestations of primary hepatic angiosarcoma. Dig Dis Sci. 2003;48:677–682. doi: 10.1023/a:1022868221670. [DOI] [PubMed] [Google Scholar]
- 23.O’Grady JG. Treatment options for other hepatic malignancies. Liver Transpl. 2000;6(Suppl 2):S23–29. doi: 10.1053/jlts.2000.18687. [DOI] [PubMed] [Google Scholar]
- 24.Zeng D, Cheng J, Gong Z, Chen J, Long H, Zhu B. A pooled analysis of primary hepatic angiosarcoma. Jpn J Clin Oncol. 2020;50:556–567. doi: 10.1093/jjco/hyaa017. [DOI] [PubMed] [Google Scholar]
- 25.Razumilava N, Gores GJ. Cholangiocarcinoma. Lancet. 2014;383:2168–2179. doi: 10.1016/S0140-6736(13)61903-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Forner A, Reig M, Bruix J. Hepatocellular carcinoma. Lancet. 2018;391:1301–1314. doi: 10.1016/S0140-6736(18)30010-2. [DOI] [PubMed] [Google Scholar]
- 27.Molina E, Hernandez A. Clinical manifestations of primary hepatic angiosarcoma. Dig Dis Sci. 2013;48:677–682. doi: 10.1023/a:1022868221670. [DOI] [PubMed] [Google Scholar]
- 28.Tran Minh M, Mazzola A, Perdigao F, Charlotte F, Rousseau G, Conti F. Primary hepatic angiosarcoma and liver transplantation: radiological, surgical, histological findings and clinical outcome. Clin Res Hepatol Gastroenterol. 2018;42:17–23. doi: 10.1016/j.clinre.2017.02.006. [DOI] [PubMed] [Google Scholar]
- 29.Infante PF, Petty SE, Groth DH, Markowitz G, Rosner D. Vinyl chloride propellant in hair spray and angiosarcoma of the liver among hairdressers and barbers: case reports. Int J Occup Environ Health. 2009;15:36–42. doi: 10.1179/107735209799449699. [DOI] [PubMed] [Google Scholar]
- 30.Palmerini E, Maki RG, Staals EL, Alberghini M, Antonescu CR, Ferrari C, Ruggieri P, Mavrogenis A, Bertoni F, Cesari M, Paioli A, Marchesi E, Picci P, Ferrari S. Primary angiosarcoma of bone: a retrospective analysis of 60 patients from 2 institutions. Am J Clin Oncol. 2014;37:528–534. doi: 10.1097/COC.0b013e31827defa1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Yuan WH, Li AF, Hsu HC, Hu YS, Lee RC. Initial clinical radiological findings and staging to predict prognosis of primary hepatic angiosarcoma: a retrospective analysis. PLoS One. 2019;14:e0225043. doi: 10.1371/journal.pone.0225043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Afrit M, Nasri M, Labidi S, Mejri N, El Benna H, Boussen H. Aggressive primary hepatic epithelioid hemangioendothelioma: a case report and literature review. Cancer Biol Med. 2017;14:187–190. doi: 10.20892/j.issn.2095-3941.2016.0105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Hu HJ, Jin YW, Jing QY, Shrestha A, Cheng NS, Li FY. Hepatic epithelioid hemangioendothelioma: dilemma and challenges in the preoperative diagnosis. World J Gastroenterol. 2016;22:9247–9250. doi: 10.3748/wjg.v22.i41.9247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Weiss SW, Enzinger FM. Epithelioid hemangioendothelioma: a vascular tumor often mistaken for a carcinoma. Cancer. 1982;50:970–981. doi: 10.1002/1097-0142(19820901)50:5<970::aid-cncr2820500527>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
- 35.Ishak KG, Sesterhenn IA, Goodman ZD, Rabin L, Stromeyer FW. Epithelioid hemangioendothelioma of the liver: a clinicopathologic and follow-up study of 32 cases. Hum Pathol. 1984;15:839–852. doi: 10.1016/s0046-8177(84)80145-8. [DOI] [PubMed] [Google Scholar]
- 36.Cardinal J, de Vera ME, Marsh JW, Steel JL, Geller DA, Fontes P, Nalesnik M, Gamblin TC. Treatment of hepatic epithelioid hemangioendothelioma a single-institution experience with 25 cases. Arch Surg. 2009;144:1035–1039. doi: 10.1001/archsurg.2009.121. [DOI] [PubMed] [Google Scholar]
- 37.Arai J, Shimozuma Y, Otoyama Y, Sugiura I, Nakajima Y, Hayashi E, Kajiwara A, Omori R, Uozumi S, Miyashita M, Uchikoshi M, Doi H, Sakaki M, Wang T, Eguchi J, Ito T, Kurihara T, Munechika J, Gokan T, Saito K, Miura S, Tate G, Takimoto M, Yoshida H. Three cases of histologically proven hepatic epithelioid hemangioendothelioma evaluated using a second-generation microbubble contrast medium in ultrasonography: case reports. BMC Gastroenterol. 2019;19:187. doi: 10.1186/s12876-019-1113-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Mehrabi A, Kashfi A, Fonouni H, Schemmer P, Schmied BM, Hallscheidt P, Schirmacher P, Weitz J, Friess H, Buchler MW, Schmidt J. Primary malignant hepatic epithelioid hemangioendothelioma: a comprehensive review of the literature with emphasis on the surgical therapy. Cancer. 2006;107:2108–2121. doi: 10.1002/cncr.22225. [DOI] [PubMed] [Google Scholar]
- 39.Gündoğan BD, Çıtak EÇ, Sağcan F, Esen K, Yıldız A, Arpacı RB. Temporal bone hemangioendothelioma as a rare vascular tumor in childhood: case report and review of the literature. Turk J Pediatr. 2020;62:843–850. doi: 10.24953/turkjped.2020.05.018. [DOI] [PubMed] [Google Scholar]
- 40.Singh AN, Kilambi R, Das P, Madhusudhan KS, Pal S. Malignant hemangiopericytoma of the liver masquerading as hepatocellular carcinoma. Indian J Surg Oncol. 2018;9:256–259. doi: 10.1007/s13193-018-0734-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Enzinger FM, Smith BH. Hemangiopericytoma. An analysis of 106 cases. Hum Pathol. 1976;7:61–82. doi: 10.1016/s0046-8177(76)80006-8. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.