Study reference

Study characteristics

Patient characteristics

Candidate predictors

Model development, performance and evaluation

Outcome measures and results

Comments

Interpretation of model

Kattan, 2002

Development MSKCC model

Source of data and date: prospective cohort, July 1982 through May 2000

Setting/ number of centres and country: single institution, NY, USA

Funding and conflicts of interest:

Supported in part by grant no. RPG-00-202-01-CCE (to M.W.K.) from the American Cancer Society and grant no. P0-CA-47179-11 (to M.F.B.) from the National Cancer Institute.

COI not reported.

Recruitment method: consecutive

Inclusion criteria:

Adult patients (> 16 years of age) who underwent treatment for primary soft tissue sarcoma at Memorial Sloan-Kettering Cancer Center.

Exclusion criteria:

Patients who presented with local or systemic recurrence were excluded from this study.

Treatment: All patients were treated with surgical resection. Some patients received adjuvant chemotherapy or radiation at the discretion of the multidisciplinary soft tissue sarcoma group or as part of clinical trials. Because treatment was not prospectively randomized but included both patients prospectively randomized in trials and those given standard of care based on prognosis, treatment variables were omitted from modeling.

Participants:

N= 2,136

Mean age:

50.9 years

Sex: % M / % F

Not reported.

Age:
Age at diagnosis

Tumor size:

≤5, 5 to 10, or > 10 cm

Histologic grade:

High or low

Histologic subtype: fibrosarcoma, leiomyosarcoma, liposarcoma, malignant fibrous histiocytoma, malignant peripheral nerve

tumor, synovial, or other.

Tumor depth:

superficial or deep

Tumor site:

Upper extremity, lower extremity, visceral, thoracic or trunk, retro intraabdominal,

or head or neck.

Missing data:

Patients whose sarcoma site was skin (n=25) were excluded. Patients with one or more missing values (n=139) were omitted, leaving 2,163 patients for analysis.

Development

Modelling method: Three nomogram development approaches were compared: Kaplan-Meier, recursive partitioning, and Cox regression.

The Cox regression model was used to develop the nomogram.

Performance

Calibration measures:

‘excellent’ calibration according to authors, shown in calibration plot.

Discrimination measures and 95%CI:

C-index: 0.77

Classification measures:

Not reported.

Evaluation

Method for testing model performance: internal.

Type of outcome: single

Definition and method for measurement of outcome:

Disease-specific survival rates, death from sarcoma or treatment complication was considered an event.

Endpoint or duration of follow-up:

Until death, maximum follow-up  18.1 years

Number of events /outcomes:

The median follow-up overall and for the patients still alive was 3.2 and 4.0 years; the 5- and 10-year disease-specific death probabilities were 25% (95% CI, 23% to 27%) and 35% (95% CI, 32% to 38%) respectively.

RESULTS

Multivariable model:

Age at diagnosis

Tumor size (< 5, 5 to 10, or > 10 cm)

Histologic grade (high or low), in Mariani 2005 changed to FNCLCC-grade (1-3)

Histologic subtype (fibrosarcoma, leiomyosarcoma, liposarcoma, malignant fibrous histiocytoma, malignant peripheral nerve tumor, synovial, or other)

Depth (superficial or deep)

Site (upper extremity, lower extremity, visceral, thoracic or trunk, retro intraabdominal, or head or neck)

No effect sizes reported.

Interpretation: confirmatory.

Authors’ conclusion

In conclusion, the nomogram estimates the probability that the patient will die of sarcoma within 12 years, assuming he or she does not die of another cause first. Such probability estimates may be useful for patient counseling, follow-up scheduling, and clinical trial eligibility determination.

Eilber, 2004

MSKCC, external validation Kattan 2002

Source of data and date: prospectively recorded hospital data, between 1975 and 2002.

Setting/ number of centres and country: department of surgery, University of California–Los Angeles (UCLA; Los Angeles, CA)

Funding and conflicts of interest:

“Supported by National Institutes of Health Program Project Grant P01CA47179 (M.F.B.), a Kristen Ann Carr Fellowship (F.C.E.), and American Cancer Society Grant RPG-00-202-01-CCE (M.W.K.).”

COI not reported.

Recruitment method: consecutive

Inclusion criteria: patients who underwent treatment for primary STS at UCLA.

Exclusion criteria: Patients who presented with locally recurrent or metastatic disease were excluded from the analysis.

All patients with STS who were treated with an ifosfamide-based chemotherapy protocol (n = 238 between 1990 and 2002) were excluded, due to evidence that ifosfamide-based chemotherapy is associated with improved survival in patients with high-risk primary extremity STS.

Treatment: All patients had their primary tumors completely surgically resected at UCLA. A significant number of patients received adjuvant radiation therapy and/or adjuvant chemotherapy. Adjuvant therapy was administered at the discretion of the multidisciplinary sarcoma research group or as part of a clinical trial.

Participants:

929 patients

Mean age:

49 years

Sex: % M / % F

NR

Other important characteristics:

Tumor grade

Low: 272 (29%)

Intermediate: 200 (21%)

High: 457 (50%)

N/A (external validation only)

Development

N/A

Performance

Calibration measures and 95%CI: calibration plots reported for nomogram with and without patients with intermediate grade disease. Model is considered to be very well calibrated according to the authors.

Discrimination measures and 95%CI:

C-index 0.76

Classification measures: NR

Evaluation

Method for testing model performance: separate external validation

Type of outcome: single

Definition and method for measurement of outcome: 12-year disease specific survival. Disease-specific survival was defined as the time from surgery to death caused by disease or to last follow-up.

Endpoint or duration of follow-up: NR.

Number of events/outcomes:

With median follow-up periods of 48 months for all patients and 60 months for surviving patients, the observed 5-year and 10-year disease-specific survival rates were 77% (95% CI, 74– 80%) and 71% (95% CI, 67–75%), respectively.

RESULTS

Multivariable model:

MSKCC model from Kattan 2002 used.

Interpretation: confirmatory.

Authors’ conclusion

In conclusion, the MSKCC Sarcoma Nomogram was found to yield accurate survival predictions when applied to an external cohort consisting of patients who were treated at UCLA.

Mariani, 2005

MSKCC adaptation model Kattan 2002

Source of data and date:

Data from institute, between January 1980 and December 2000

Setting/ number of centres and country: the Istituto Nazionale

per lo Studio e la Cura dei Tumori (INT) (Milan, Italy).

Funding and conflicts of interest: NR

Recruitment method: consecutive

Inclusion criteria:

patients with localized extremity STS underwent

surgery with curative intent, who presented with primary disease

Exclusion criteria:

-

Treatment:

All surgical resections were macroscopically complete, which we defined as the absence of macroscopic residual disease after surgical excision of the tumor. Adjuvant radiation therapy was delivered to 237 patients (37%). External beam radiation was used in all such patients, and the doses ranged from 45 grays (Gy) to 65 Gy (median, 57 Gy). Adjuvant chemotherapy (mainly anthracycline-based regimens associated with ifosfamide) was given to 114 patients (18%) at the discretion of the multidisciplinary STS group or as part of clinical trials.

Participants:

642 patients

Mean age:

47.7 years

Sex: % M / % F

52/48

Other important characteristics:

Histologic grade:

Grade 1: 180 (28%)

Grade 2: 170 (26%)

Grade 3: 292 (46%)

Predictors same as MSKCC model Kattan 2002, only histologic grade 1-3 instead of high vs low.

Missing data: NR

Development

Modelling method: For MSKCC model testing and revision, we adopted the approach of “validation by calibration”, Cox model.

Performance

Calibration measures and 95%CI:

“Graphic comparison of observed and predicted sarcoma-specific survival curves showed that predictions by the nomogram were quite accurate, within 10% of actual survival for all prognostic strata. Statistical analysis showed that such predictions could be improved by employing approximately 25% shrinkage to achieve good calibration”

Discrimination measures and 95%CI:

C-statistic: 0.76

Classification measures: NR

Evaluation

Method for testing model performance:

“To account for possible over fitting, we calculated the degree of shrinkage of Cox model regression coefficients and the optimism in the estimated c statistic by means of bootstrap”

Type of outcome: single

Definition and method for measurement of outcome:

10-year extremity STS-specific death: “Survival time, which was computed from the date of surgery to the date of death or last follow-up, was censored for living patients and for patients who died of causes unrelated to STS, because we modeled disease-specific death.”

Endpoint or duration of follow-up: 120 months

Number of events/outcomes:

There were 176 deaths overall; of these, 143 deaths (81%) were due to sarcoma and, thus, contributed to the current analysis.

RESULTS

Multivariable model:

Only HR reported for adjusted predictor.

Histologic grade

Grade 2 vs. Grade 1 HR 4.51 (95% CI 1.99 to 10.2)

Grade 3 vs. Grade 1 HR 8.93 (95%CI 4.14 to 19.3)

Interpretation: confirmatory

Authors’ conclusion

In conclusion, the current study confirmed that the MSKCC nomogram is a valuable tool for individual prognostic assessment. However, some degree of adjustment seems useful for improving the quality of predictions. This hypothetically may reflect either statistical “over fitting” in the original model, weaker prognostic effect of covariates in extremity STS compared with STS in other sites, the application of a three-grade system instead of two-grade system, or some combination of the above mechanisms. The revised nomogram incorporates such an adjustment of predictions, and it is proposed as an extension in extremity STS of the MSKCC nomogram whenever histologic grade is classified according to the FNCLCC system, which is now the system used most widely all over the world.

Squires 2022

External validation MSKCC (Mariani 2005) / Sarculator (Callegaro 2016)

Source of data and date: U.S. Sarcoma Collaborative (USSC) database, from 2000 to 2017

Setting/ number of centres and country: nine high-volume academic institutions across the United States

Funding and conflicts of interest:

FUNDING This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

DISCLOSURES The authors have no financial or conflict of interest disclosures.

Recruitment method: consecutive

Inclusion criteria: all patients who underwent

resection of primary extremity STS. Patients aged 18 years or older who underwent curative intent resection of primary extremity

STS were included.

Exclusion criteria:

Histologies excluded

from the original Sarculator nomogram development study were also excluded in the current analysis: desmoid fibromatosis, peripheral primitive neuroectodermal tumor (PPNET), alveolar or embryonal rhabdomyosarcoma, dermatofibrosarcoma protuberans, and well-differentiated liposarcoma.

Patients with metastatic or recurrent

disease were excluded.

Treatment:

All patients underwent curative intent resection of primary extremity STS.

Perioperative

Chemotherapy (n=313 (24%)) and radiation (n=700 (53%)) data also were collected.

Participants:

N=1,326

Median age [IQR]: 59 [46–71]

Sex: % M / % F

54/46

N/A (external validation only)

Development

N/A

Performance

Calibration measures:

Calibration plots: The calibration plots showed good predictability according to the authors for 5- and 10-year OS using the Sarculator nomogram.

The calibration plots for DSS demonstrated similarly good calibration using the MSKCC nomogram.

Discrimination measures and 95%CI:

Sarculator: The C-indices for 5- and 10-year OS were 0.72 (95% CI: 0.70–0.75) and 0.73 (95% CI: 0.70–0.75).

MSKCC: C-indices for 4-, 8-, and 12-year of 0.71 (95% CI: 0.68–0.75)

Classification measures:

NR

Evaluation

Method for testing model performance:

Type of outcome: single

Definition and method for measurement of outcome:

Sarculator: overall survival

MSKCC: disease-specific survival

Endpoint or duration of follow-up: NR

Number of events/outcomes:

Median follow-up time was 34 months. Median OS was 173 months (IQR, 128 months-MNR), with estimated 5- and 10-year OS of 70% and 58%,

respectively.

RESULTS

Multivariable model:

N/A (external validation)

Alternative presentation of final model:

N/A (external validation)

Interpretation: confirmatory.

Authors’ conclusion:

In conclusion, the Sarculator and MSKCC nomograms were both found to have good discriminative and prognostic ability within a diverse, modern, multi-institutional U.S. validation cohort of patients undergoing resection of primary extremity STS. Ongoing incorporation of these prognostic nomograms into the clinical management of extremity STS patients appears warranted.

Sampo, 2012

Development and external validation SAM-model

Source of data and date:

Patients referred during 1987–2002

Swedish database: 25-year period 1973–1997

Setting/ number of centres and country:

Helsinki University Central Hospital, Finland, external validation from Lund University Hospital register, Sweden

Funding and conflicts of interest:

The study was supported by the Helsinki University Central Hospital Research Funds, Finnish Cancer Society, and the Sigrid Juselius Foundation. Dr M Sampo was supported by grants from the K Albin Johansson Foundation, Finska Läkaresällskapet, and Duodecim Foundation.

COI not reported.

Recruitment method: consecutive

Inclusion criteria:

All patients referred for non-metastatic, primary or locally

recurrent STS of the extremities or trunk wall to the Soft Tissue Sarcoma Group between August 1987 and December 2002 are included.

Exclusion criteria:

Exclusion criteria comprised: extra skeletal osteosarcoma,

chondrosarcoma, Ewing/ PNET family tumour, angiosarcoma, alveolar soft tissue sarcoma, epithelioid sarcoma, clear cell sarcoma,

atypical lipoma/grade I liposarcoma, dermatofibrosarcoma protuberans or preoperative radiation therapy. A total of 15 patients with chemotherapy were also excluded.

Treatment:

The primary treatment in all cases was a surgical resection. If the preoperative investigations indicated that adequate surgical margins were not achievable, surgery aimed at marginal surgical margins with postoperative radiation therapy. The treatment protocol recommended, following intralesional surgery, a reoperation when feasible.

Participants:

N=294

Validation database, N=354

Mean age (range)

57 (16-92)

Validation database 63 (17-96)

Sex: % M / % F

52/48

Validation database: 56/44

Necrosis:

Absent or present.

Vascular invasion:

Absent or present.

Tumor size:

In cm, recorded as the largest diameter of tumor in the surgical specimen reported by the original pathologist.

Histological grade:

The pathologist  assigned the histological malignancy grade of the tumor based on a four-tiered grading scale modified from Broders et al (1939) and

Angervall et al (1986). Grades 1 and 2 are low grades and 3 and 4 high grades.

Tumor depth:

Subcutaneous tumors with or without cutaneous

extension but without involvement of the deep fascia were defined superficial, all others deep.

Tumor location:

Extremity or axis of body

Missing data:

In 84 cases, we were unable to retrieve the original histological slides leaving 294 tumours to analysis. Demographic data for

missing cases was similar except for histological subtype.

Development

Modelling method: Cox regression multivariate model

Performance

Calibration measures and 95%CI:

Calibration plots reported: “A good concordance is seen in the groups with a predicted 10-year survival of over 50%, whereas a slight underestimation is observed in the groups predicted to have the lowest survival.”

Discrimination measures and 95%CI:

AUC: 0.81 (95% CI 0.75–0.87)

C-index: 0.79

Validation series:

AUC: 0.77 (95% CI 0.72–0.82)

C-index: 0.77

Classification  measures:

Compared to SIN model: when the patients were classified into three categories (cutoff at tertiles) on the basis of their predicted 10-year sarcoma-specific survival, the net reclassification improvement (NRI 0.12, P=0.03) is significant as well as the integrated discrimination improvement (IDI 0.03, P=0.0003

Evaluation

Method for testing model performance: external

Type of outcome: single

Definition and method for measurement of outcome:

Sarcoma-specific survival

(SSS) was calculated from the date of the diagnosis to death from sarcoma. Deaths due to other causes than sarcoma were censored.

Endpoint or duration of follow-up: Until death.

Number of events/outcomes:

The median follow-up for the patients alive at the end of follow-up was 7.2 years (range 0.3–17.5 years). The 5-year sarcoma-specific survival rate was 75% (95% CI 0.70–0.80) and at 10 years 71% (95% CI 0.64–0.76)

RESULTS

Multivariable model:

Tumor size per cm: HR 1.10 (1.05 to 1.15)

Necrosis (no/yes): HR 1.60 (0.88 to 2.90)

Vascular invasion (no/yes): HR 1.60 (0.93 to 2.75)

Histological grade (2/3/4, per grade): HR 1.57 (1.11 to 2.22)

Tumor depth (superficial/ deep): HR 3.51 (1.71 to 7.38)

Location (extremity/ axis of body): HR 1.65 (1.01 to 2.68)

Interpretation: exploratory

Authors’ conclusion

In conclusion, we have created a new prognostic model to

estimate survival probability in patients with the commonest

subtypes of STS. An external validation was performed showing a good prognostic accuracy and an improvement in accuracy compared with a model with size, necrosis, and vascular invasion only. Our model can be seen as a working formulation to be

refined by validation in further external validation studies and is made available online.

Callegaro, 2016

Development  and external validation Sarculator model

Source of data and date:

Development cohort: 1 Jan 1994, to 31 Dec 2013

External validation, cohort 1: 1 Jan 1996 to 15 May 2012, cohort 2: 1 Jan 1994 to 31 Dec 2013, cohort 3: 1 Jan 2006 to 31 Dec 2013

Setting/ number of centres and country:

Development cohort: Istituto Nazionale Tumori (Milan, Italy).

Validation cohorts: Institut Gustave

Roussy (Villejuif, France), Mount Sinai Hospital (Toronto, ON, Canada), Royal Marsden Hospital (London, UK)

Funding and conflicts of interest:

The authors declare no competing interests.

Funding: None

Recruitment method:

Consecutive

Inclusion criteria:

All consecutive adult (aged >18 years) patients with primary (non-recurrent and non-metastatic) soft-tissue sarcomas of the extremities, who had had an operation with curative intent at Fondazione IRCCS Istituto Nazionale dei Tumori (Milan, Italy), between Jan 1, 1994, and Dec 31, 2013, formed the development cohort of the study. We defined soft-tissue sarcomas of the extremities as all tumours arising from the shoulder girdle to the hand (upper extremity) and from the pelvic girdle (excluding endopelvic tumours) to the foot (lower extremity).

Exclusion criteria:

We excluded patients with desmoids, soft-tissue Ewing’s sarcoma, alveolar or embryonal rhabdomyosarcoma, dermatofibrosarcoma protuberans, and well differentiated liposarcoma because of the peculiar natural histories and treatment strategies for these cancers.

Treatment:

The indication to administer radiotherapy was given by both the operating surgeon and the radiation oncologist when a higher risk of relapse was thought to exist based on clinical grounds. However, no prospectively selected criteria were used to this end. Chemotherapy was given at the discretion of the  multidisciplinary institutional sarcoma board or as part of ongoing clinical trials.

Participants:

N Development cohort (DC): 1,452

N Validation cohort (VC)1: 420

N VC2: 1,436

N VC3: 444

Median age (IQR):

DC: 54 (40-66)

VC1: 51 (38-62)

VC2: 57 (43-70)

VC3: 63 (50-74)

Sex: % M / % F

DC: 54/46

VC1: 51/49

VC2: 56/44

VC3: 57/43

Age at diagnosis:

In years.

Tumor size:

In cm.

Tumor depth:

Superficial or deep according to the investing fascia.

Surgical margins:

We classified all macroscopically complete resections according to the closest surgical margin, which we microscopically categorised as either positive (tumour within 1 mm from the inked surface; R1) or negative (absence of tumour within 1 mm from the inked surface; R0). We excluded macroscopically incomplete resection from the analysis.

Tumor grading:

Fédération Française des Centres de Lutte Contre le Cancer (FNCLCC; French Federation of Centers for the Fight against Cancer) Criteria, grades I, II, and III.

Histological subtypes:

Based on WHO’s criteria and grouped into nine categories: leiomyosarcoma, dedifferentiated or pleomorphic liposarcoma, myxoid liposarcoma, malignant peripheral nerve sheath tumours, myxofibrosarcoma, synovial sarcoma, undifferentiated pleomorphic sarcoma, vascular sarcoma (including both epithelioid haemangio-endothelioma [mostly grade 1 and occasionally grade 2] and angiosarcoma [only grade 3]), and others.

Number of participants with any missing value?

Not reported.

How were missing data handled?

Not reported.

Development

Modelling method: Multivariable Cox model, backward selection.

Performance

Calibration measures and 95%CI:

Well-calibrated according to authors. Calibration plot, Hosmer–Lemeshow calibration test reported.

Discrimination measures, C-index (95% CI):

DC: 0.767 (0.743 to 0.789).

VC1: 0.698 (0.638 to 0.754)

VC2: 0.775 (0.754 to 0.796)

VC3: 0.762 (0.720 to 0.806)

Classification measures:

Not reported.

Evaluation

Method for testing model performance: internal and external

Type of outcome: single

Definition and method for measurement of outcome:

Overall survival (events: deaths from any cause)

Endpoint or duration of follow-up:

The median follow-up was 86 months (IQR 47–123) for the development cohort; 75 months (46–117) for the French validation cohort, 85 months (44–121) for the Canadian validation cohort, and 54 months (30–71) for the UK validation cohort

Number of events/outcomes:

In the development cohort, overall survival was 79.9% (95% CI 77.7–82.1) at 5 years and 72.9% (70.2–75.7) at 10 years follow-up. In the validation cohorts, 5-year and 10-year overall survival were 78.1% (95% CI 73.9–82.6) and 68.3% (62.6–74.5) for French patients; 72.7% (70.2–75.2) and 60.2% (57.0–63.5) for Canadian patients; and 72.7% (68.1–77.5) and not estimated (due to the shorter follow-up) for the UK patients.

RESULTS

Multivariable model, HR (95% CI):

Age

66 years vs 40 years: 1.58 (1.30–1.93)

Tumour size

10 cm vs 4 cm: 2.48 (1.92–3.21)

FNCLCC grade

II vs I 2.68 (1.64–4.39)

III vs I 4.25 (2.64–6.84)

Histological subtype

Leiomyosarcoma vs myxoid

Liposarcoma: 2.50 (1.51–4.16)

DD/pleom lipo vs myxoid liposarcoma: 1.48 (0.80–2.74)

MPNST vs myxoid liposarcoma: 1.89 (1.06–3.36)

Myxofibrosarcoma vs myxoid

Liposarcoma: 1.64 (0.99–2.70)

Synovial vs myxoid liposarcoma: 2.70 (1.59–4.60)

UPS vs myxoid liposarcoma: 1.27 (0.76–2.11)

Vascular vs myxoid liposarcoma: 5.81 (2.71–12.45)

Other vs myxoid liposarcoma: 1.99 (1.23–3.21)

Alternative presentation of final model: Nomogram, free app called Sarculator has been developed

for smartphones and tablets and is distributed via the official app stores

Interpretation: confirmatory

Authors’ conclusion

Our nomograms are reliable prognostic methods that can be used to predict overall survival and distant metastases in patients after surgical resection of soft-tissue sarcoma of the extremities. These nomograms can be offered to clinicians to improve their abilities to assess patient prognosis, strengthen the prognosis-based decision making, enhance patient stratification, and inform patients in the clinic.

It is important to note that these nomograms only apply to adult patients with primary soft-tissue sarcomas of the extremities, who underwent macroscopically complete surgical resection at multidisciplinary sarcoma centres.

Callegaro 2019

Development and external validation of dynamic Sarculator model

Source of data and date, setting/ number of centres and country:

All consecutive adult (>18years) patients with primary (non-recurrent, non-metastatic) eSTS surgically treated at Fondazione IRCCS Istituto Nazionale dei Tumori (Milan, Italy), Institut Gustave Roussy (Villejuif, France), Mount Sinai Hospital (Toronto, Canada), and at the Royal Marsden Hospital (London ,UK) from 1994 to 2013 were merged, forming the development cohort. On the Milan series, we developed two static nomograms for OS and DM in 2016. Patients with the same characteristics operated on between 2000 and 2016 at 7 other European referral centers comprised the validation cohort

Funding and conflicts of interest:

Recruitment method: consecutive

Inclusion criteria:

Adult (>18years) patients with primary (non-recurrent, non-metastatic) eSTS surgically treated. Extremity STS were defined as tumors arising between the shoulder girdle and the hand (upper extremity) and between the pelvic girdle (excluding endopelvic tumours) and the foot (lower extremity).

Exclusion criteria:

Patients with well-differentiated liposarcoma, dermatofibrosarcoma protuberans, desmoid-type fibromatosis, Ewing sarcoma and alveolar or embryonal rhabdomyosarcoma were excluded.

Patients who underwent macroscopically incomplete (R2) resections were excluded.

Treatment:

Patients were operated with curative intent. Radiotherapy (RTx) and/or chemotherapy (CTx) were used according to multidisciplinary guidance or as part of clinical trials. 

Participants:

N development cohort (DC): 3,740

N validation cohort (VC): 893

Median age (IQR):

DC: 56 (42–69)

VC: 62 (49–73)

Sex: % M / % F

DC: 54.8/45.2

VC: 55.3/44.7

Predictors (candidate & selected):

In the multivariable Cox landmark OS supermodel, after application of the backward procedure the following variables were excluded from the covariates set: tumor's depth, surgical margin status, CTx administration, RTx administration.

The final supermodel included age at surgery, tumor size and its interaction with T_LM, grading and its interaction with T_LM, histology, and both LR and DM indicator variables. 

Development

Modelling method: The dynamic nomogram was developed using a landmark analysis approach and a multivariable Cox model. A backward procedure based on the Akaike Information Criterion was adopted for variable selection.

Performance

Calibration measures:

Calibration plots were reported, good calibration according to authors.

Discrimination measures and 95%CI:

In the development series, the Harrell C index was (95% bootstrap confidence interval) 0.776 (0.761–0.790) for predictions calculated at time of primary surgery (T_LM=0) and 0.837 (0.822–0.851), 0.845 (0.823–0.862) and 0.834 (0.811–0.859) for predictions calculated at 1 year, 2 years and 3 years after surgery, respectively.

In the validation series, the Harrell C index was 0.675 (0.643–0.704) at T_LM=0, 0.773 (0.740–0.801) at T_LM=12 months, 0.810 (0.775–0.844) at T_LM=24 months and 0.796 (0.751–0.834) at T_LM=36 months.

Classification measures:

NR

Evaluation

Method for testing model performance:

Internal and external

Type of outcome: single

Definition and method for measurement of outcome:

5-year overall survival at different times during the first three years of follow-up.

Endpoint or duration of follow-up: NR

Number of events/outcomes:

The median follow-up was (interquartile [IQ] range) 79 months (44–119 months) for the development cohort and 71 months (43–108 months) for the validation cohort. In the development and validation cohorts, respectively, 1003 and 367 patients died. In the development cohort, 5-year OS was 76.0% (74.6–77.5%) and 10-year OS was 66.3% (64.3–68.2%). In the validation cohort 5- and 10-year OS was 59.5% (56.0–63.1%) and 48.0% (43.8–52.6%), respectively. 

RESULTS

Multivariable model:

Covariates: HR (95% CI)

Age, years               

 69 vs. 42: 1.80 (1.58,2.05)               

Local recurrence     

 yes vs. no: 5.63 (4.26,7.44)

Distant Metastasis   

 yes vs. no: 10.34 (8.74,12.23)

Histological subtype

 LMS vs. Myxoid lipo: 1.78(1.26,2.52)

 DD/pleom lipo vs. Myxoid lipo: 1.37 (0.93,2.02)

 MPNST vs. Myxoid lipo: 1.73 (1.16,2.58)       

 Myxofibro vs. Myxoid lipo: 1.05 (0.72,1.53)

 Synovial sarcoma vs. Myxoid lipo: 2.03 (1.43,2.88)               

 UPS vs. Myxoid lipo:1.18 (0.85,1.63)

 Vascular vs. Myxoid lipo: 3.20 (1.85,5.53)

 Other vs. Myxoid lipo: 1.48 (1.07,2.04)

Size, cm  

 11 vs. 4 (0): 3.06 (2.53,3.70)

 11 vs. 4 (12): 2.32 (1.92, 2.80)

 11 vs. 4 (24): 1.90 (1.55, 2.32)

 11 vs. 4 (36): 1.65 (1.29, 2.11)

FNCLCC grade         

 II vs. I (0): 2.55 (1.75, 3.73)

 II vs. I (12): 2.07 (1.42, 3.01)

 II vs. I (24): 1.63 (1.11, 2.40)

 II vs. I (36): 1.26 (0.82, 1.94)

 III vs. I (0): 4.88 (3.40,7.02)

 III vs. I (12): 2.59 (1.79,3.75)

 III vs. I (24): 1.59 (1.08, 2.33)

 III vs. I (36): 1.09 (0.72,1.67)               

Alternative presentation of final model: dynamic nomogram. The new nomogram has also been incorporated in the app ‘Sarculator’ for smartphones and tablets, which is available for free download.

Interpretation: confirmatory

Authors’ conclusion

In conclusion, this new prognostic tool fulfills a need of the oncologist dealing with eSTS patients: being able to objectively counsel patients regarding their personalized residual risk during FU. Patients might be comforted from an improvement in prognosis as the time goes by without events and the update of the prognostic estimate may also support patients’ planning for the future. Moreover, the dynamic prediction informs the physician of how a postoperative event will impact on an individual patient's prognosis quantitatively. Finally, this study paves the way for future FU personalization with possible creation of risk-adapted FU strategies. 

Voss 2022

External validation Sarculator

Source of data and date: the National Cancer Data Base (NCDB) Sarcoma

Participant Use File (PUF) between 2004 and 2017.

Setting/ number of centres and country:

The NCDB is a prospectively maintained, national, hospital-based registry that includes data from patients accounting for more than 70% of incident cancer diagnoses at over 1500 Commission on Cancer (CoC)-accredited centers in the USA.

Funding and conflicts of interest:

DISCLOSURES None.

Recruitment method: consecutive

Inclusion criteria: Patients with soft tissue sarcoma of the extremity or trunk from the National Cancer Data Base (NCDB) Sarcoma Participant Use File (PUF) between 2004 and 2017

were included.

Briefly, we included extremity and trunk sites (ICD-O-3 topography codes C471, C472, C476, C491, C492, and C496) with stage I–III disease by AJCC 8^th edition staging. The following histologies were included on the basis of their inclusion in the original Sarculator algorithm (ICD-O histology codes in parentheses): leiomyosarcoma (8890, 8891, 8896), liposarcoma [8850, 8855, 8857 (grades 2 and 3 only)], dedifferentiated liposarcoma [8858 (grades 2 and 3 only)], pleomorphic liposarcoma (8854), myxoid liposarcoma (8852–53), malignant peripheral nerve sheath tumor (8540, 8561), myxofibrosarcoma (8840), synovial sarcoma (9040–43), vascular sarcomas (angiosarcoma 8894, 9120; hemangioendothelioma 9130, 9133), undifferentiated pleomorphic sarcoma (8805, 8830), or other sarcoma (8000–01, 8004, 8800–01, 8804, 8810–11, 8813, 8815, 8825, 8895, 9044, 9150, 9170, 9364, 9580, 9581).

We included only patients who underwent surgery and had either an R0 (no residual tumor at the primary site) or R1 (microscopic residual tumor) resection as Sarculator was only designed for those who have undergone complete surgical resection.

Exclusion criteria:

We excluded those with incomplete grade, treatment, or survival data; those with metastatic disease; and those with a tumor <1 cm or > 35 cm in size (maximal size accepted by Sarculator is 35 cm).

Treatment: All patients underwent complete surgical resection.

Radiation therapy:

Neoadjuvant: n=1,941 (19.93%)

Adjuvant: 3,856 (39.60%) None: 3,941 (40.47%)

Chemotherapy

Neoadjuvant or adjuvant: 1,572 (16.14%)

Participants:

N= 9,738

Age: N(%)

• <50: 2,827 (29.03)
• 50–59: 1,916 (19.68)
• 60–69: 1,999 (20.53)
• 70–79: 1,720 (17.66)
• ≥ 80: 1,276 (13.10)

Sex: % M / % F

54.10/45.90

N/A (external validation only)

Development

Modelling method:

N/A

Performance

Calibration measures:

Calibration plots: Sarculator tends to slightly overestimate survival for the higher survival quintiles and tends to underestimate the survival for the subgroup with the lowest actual OS

Discrimination measures and 95%CI:

C-index of 0.726

Classification measures:

NR

Evaluation

Method for testing model performance:

External validation

Type of outcome: single

Definition and method for measurement of outcome:

Overall survival

Endpoint or duration of follow-up: NR/until death

Number of events/outcomes:

mean follow-up time of 4.45 years. The 5-year actual OS for the study cohort was 68.9%.

RESULTS

Multivariable model: N/A (external validation only)

Alternative presentation of final model: N/A (external validation only)

Interpretation: confirmatory.

Authors’ conclusion

Sarculator is overall a good predictor of aOS and useful tool for clinicians to aid in survival prognostication, but clinicians should be aware of subpopulations for whom Sarculator’s predictions may be stronger or more limited. Future work may focus on enhancing the Sarculator algorithm specifically for US patients, including the incorporation of predictive demographic variables.

Van Praag, 2017

Development and internal validation PERSARC model

Source of data and date: retrospective cohort, January 2001 – December 2014

Setting/ number of centers and country: multicenter study, five international

sarcoma centers

Conflict of interest statement: None declared.

Funding: This study was supported by the Dutch Cancer Society - KWF Kankerbestrijding.

Role of the funding source: This funding source had no role in the design of this study as well as any role during its execution, analyses, interpretation of the data, in the writing of the report or decision to submit the article for publication.

Recruitment method: consecutive series of patients who underwent surgery

Inclusion criteria:

Eligible diagnoses included high grade (FNCLCC grade III) angiosarcoma, malignant peripheral nerve sheath tumor, synovial sarcoma, spindle cell sarcoma, myxofibrosarcoma and (pleomorphic) soft-tissue sarcomas not-otherwise-specified.

Exclusion criteria:

Excluded patients include those that were treated without curative intent, had LR or DM within 2 months after primary treatment (ruled out by pre-treatment and follow-up (FU) staging with lung computed tomography (CT) scan), had a tumor in their abdomen, thorax, head or neck or received (neo) adjuvant treatment other than RT or chemotherapy.

Treatment received?

Participants:

N= 766

Mean age ± SD:

58.28 ± 19.39

Sex: % M / % F

57 / 43

Other important characteristics:

Age:

Patient age at presentation.

Tumor size:

In cm. Maximum diameter at pathologic analysis. In patients that received neoadjuvant RT and/or chemotherapy, tumor size was

defined as maximum diameter measured by CT or MRI before treatment.

Depth:

Relative to the investing fascia: deep, superficial, deep and superficial.

Histology subtype:

Obtained from medical records:

• Myxofibrosarcoma
• MPNST
• Synovial sarcoma
• Spindle cell sarcoma
• MFH/UPS
• other

Surgical margin:

• Intralesional: for tumor cells present at the margin of the resection specimen (<0.1 mm)
• Marginal: tumor cells found within 0.1 - 2 mm of the margin a
• Free: tumor cells found at least 2 mm away from the margin

RT:

Information from medical records: No RT, neoadjuvant, adjuvant

Number of participants with any missing value?

N (%): 72 patients (8.6%) of original 838

How were missing data handled?

Patients with missing values were not included in the development of the model.

Development

Modelling method:

Outcome OS: multivariate Cox proportional hazards regression model

Outcome CILR: Fine and Gray model

Performance

Calibration measures and 95%CI: Calibration plots are reported.

Discrimination measures and 95%CI:

C-index for OS: 0.677 (95% CI 0.643 to 0.701.

C-index for LR: 0.696 (95% CI 0.629 to 0.743)

Classification measures:

NR

Evaluation

Method for testing model performance: predictive performance of the prediction

models was assessed internally by using leave-one-out cross validation (CV).

Type of outcome: Overall survival (OS), cumulative incidence of local recurrence (CILR)

Definition and method for measurement of outcome:

OS: overall survival at 3, 5 and 10 years after surgery

CILR: cumulative incidence of local recurrence in the presence of competing events. LR at 3, 5 and 10 years after surgery

Endpoint or duration of follow-up:

Patients visited the outpatient clinic for their scheduled clinical and radiographic FU: every 3-4 months in the first 2-3 years, then every 6 months and after 5 years yearly. It was common that FU was ended after 10 years evidence of no disease.

Number of events/outcomes:

OS was estimated to be equal to 63%, 53% and 39% at 3, 5 and 10 years, respectively; LR was estimated to be equal to 13.3%, 15.1% and 17.2% at 3, 5 and 10 years, respectively.

RESULTS

Multivariable model OS:

Age (unit increase of 10 years): HR 1.195 (1.116 to 1.268)

Size (unit increase of 1 cm): HR 1.068 (1.052 to 1.085)

Depth (relative to investing fascia)

Superficial vs deep: HR 0.813 (0.591 to 1.117)

Deep and superficial vs deep: HR 1.110 (0.736 to 1.674)

Histology

MPNST vs myxofibrosarcoma: HR 1.422 (0.989 to 2.044)

Synovial sarcoma vs myxofibrosarcoma: HR 1.261 (0.869 to 1.831)

Spindle cell sarcoma vs myxofibrosarcoma: HR 1.211 (0.884 to 1.661)

MFH/UPS vs myxofibrosarcoma: HR 1.293 (0.890 to 1.876)

Margin

0.1 to 0.2 mm vs 0 mm: HR 0.786 (0.599 to 1.033)

> 2 mm vs 0 mm: HR 0.711 (0.524 to 0.964)

RT

Neoadjuvant vs no RT: HR 0.548 (0.399 to 0.753)

Adjuvant vs no RT: HR 0.638 (0.486 to 0.837)

Multivariable model LR:

Age (unit increase of 10 years): sHR 1.051 (0.942 to 1.184)

Size (unit increase of 1 cm): sHR 1.031 (1.001 to 1.063)

Depth (relative to investing fascia)

Superficial vs deep: sHR 0.907 (0.536 to 1.535)

Deep and superficial vs deep: sHR 0.563 (0.198 to 1.604)

Histology

MPNST vs myxofibrosarcoma: sHR 1.079 (0.580 to 2.009)

Synovial sarcoma vs myxofibrosarcoma: sHR 0.779 (0.379 to 1.602)

Spindle cell sarcoma vs myxofibrosarcoma: sHR 0.979 (0.570 to 1.681)

MFH/UPS vs myxofibrosarcoma: sHR 1.096 (0.557 to 2.156)

Margin

0.1 to 0.2 mm vs 0 mm: sHR 0.635 (0.406 to 0.992)

> 2 mm vs 0 mm: sHR 0.282 (0.159 to 0.500)

RT

Neoadjuvant vs no RT: sHR 0.312 (0.146 to 0.668)

Adjuvant vs no RT: sHR 0.700 (0.417 to 1.175)

Interpretation: exploratory

Authors’ conclusion

In this study, we developed the PERSARC model which uniquely presents clinicians with the possibility to accurately predict outcome of OS and CILR and compare different treatment modalities, for patients with high-grade ESTS that undergo surgical resection with curative intent.

Smolle, 2019

Development and validation of dynamic PERSARC model for local recurrence

Source of data and date: prospectively maintained STS databases at 5 participating

tertiary sarcoma referral centers (2 for validation cohort), between January 1994 and October 2014 for the test cohort and between January 2000 and December 2013 for the validation cohort.

Setting/ number of centres and country: multicenter study, country NR

Funding and conflicts of interest:

Funding: This work was supported by the Dutch Cancer Society (DCS)—KWF Kankerbestrijding [UL2015-8028]. The funding source had no role in the design of this study; execution, analyses, and interpretation of the data; report writing; or decision to submit the article for publication.

Conflicts of Interest: Author van de Sande reports grants from Daiichi Sankyo, outside the submitted work. The remaining authors (Maria A Smolle, Dario Callegaro, JayWunder, Andrew J. Hayes, Lukas Leitner, Marko Bergovec, Per-Ulf Tunn, Veroniek van Praag, Marta Fiocco, Joannis Panotopoulos, Madeleine Willegger, Reinhard Windhager, Sander Djikstra, Winan J van Houdt, Jakob M Riedl, Michael Stotz, Armin Gerger, Martin Pichler, Herbert Stöger, Bernadette Liegl-Atzwanger, Josef Smolle, Dimosthenis Andreou, Andreas Leithner, Alessandro Gronchi, Rick L. Haas, and Joanna Szkandera) have no conflicts of interest to declare.

Recruitment method: consecutive

Inclusion criteria: Patients with primary nonmetastatic

high-grade (G2/3) eSTS managed with surgery at a curative intent were included in the test cohort, with patient information deriving from prospectively maintained STS databases at 5 participating tertiary sarcoma referral centers.

Extremity STS were defined as tumors from

the shoulder to the fingers (=upper limb) and from the pelvic girdle, excluding intrapelvic STS, to the foot (=lower limb).

Exclusion criteria:

Patients with missing information on oncological follow-up (i.e., development of LR/DM) had to be excluded (n = 42).

Treatment:

All patients underwent surgery at a curative intent. (Neo-)adjuvant RTX and CTX had been administered in case a high risk of LR or DM had been anticipated by the multidisciplinary tumor board, according to locally preferred guidelines, LR was defined as a radiologically and/or histologically confirmed tumor recurrence.

Participants:

Development cohort (DC) N=1,931

Validation cohort (VC) N=1,085

Median age (IQR):

DC: 59 years (44.7 to 70)

VC: 61 years (47 to 74)

Sex: % M / % F

DC: 53.8/46.2

VC: 56.7/43.3

Gender, tumor size, histological subtype (except for angiosarcoma/vascular sarcoma (p = 0.127) and dedifferentiated/ pleomorphic liposarcoma (p = 0.254), margins, neoadjuvant and adjuvant RTX, as well as adjuvant CTX (all p < 0.05) had a significant influence on risk of LR in the stepwise backward selection of the Fine and Gray model. Grading as a time-dependent effect was kept in the model (p = 0.108), while age (p = 0.082) and neoadjuvant CTX (p = 0.214) were excluded. Consequently, gender, grading, tumor size, neoadjuvant and adjuvant RTX, histological subtype, and adjuvant CTX were included in the flexible parametric competing risk regression model

Development

Modelling method:  Royston and Parmar approach to fit a flexible parametric competing risk regression model in order to estimate the risk of LR and DM, with death as the competing event; variable selection for the LR and DM models was based on a stepwise backward procedure

using a multivariable Fine and Gray model

Performance

Calibration measures and 95%CI:

The authors concluded that calibration plots for LR using test and validation cohort showed that the LR model tended to underestimate the actual patient risk, especially in the validation cohort.

Discrimination measures and 95%CI:

The Harrell C index for LR was equal to 0.705 and 0.683 for the internal and external cohort, respectively.

Classification measures:

NR

Evaluation

Method for testing model performance:

Internal and external

Type of outcome: single. (second model with outcome DM not included in present analysis)

Definition and method for measurement of outcome:

Local recurrence, defined as a radiologically and/or histologically confirmed tumor recurrence.

Endpoint or duration of follow-up: until death/NR

Number of events/outcomes: NR

RESULTS

Multivariable model:

Local Recurrence, coefficient (95% CI)

Gender    Male 1

Female 0.698 (0.529 0.921)

Grading    G2 1

G3 0.816 (0.598 1.113)

Tumor size 1.026 (1.004 1.049)

Margins    R0 1

R1/R2 2.761 (2.021 3.774)

Histology

Myxoid Liposarcoma 1

MPNST 4.227 (1.837 9.729)

Myxofibrosarcoma 4.156 (2.056 8.400)

Synovial Sarcoma 3.116 (1.429 7.014)

UPS 3.373 (1.620 7.025)

Angiosarcoma/Vascular Sarcoma 3.316 (0.981 12.341)

Dedifferentiated/Pleomorphic Liposarcoma 1.727 (0.719 4.143)

Leiomyosarcoma 2.779 (1.294 5.966)

Others 2.385 (1.123 5.065)

Neoadjuvant RTX

No 1

Yes 0.298 (0.178 0.494)

Adjuvant RTX

No 1

Yes 0.603 (0.447 0.814)

Adjuvant CTX

No 1

Yes 1.711 1.154 2.538

Restricted cubic spline 1 2.104 (1.851 2.392)

Restricted cubic spline 2 1.332 (1.230 1.442)

Restricted cubic spline 3 0.980 (0.937 1.026)

Restricted cubic spline for time-dependent effect of grading 0.944 (0.813 1.096)

Constant 0.048 (0.024 0.097)

Alternative presentation of final model: models

included in the updated version of the PERSARC app for Individualized Sarcoma Care and follow-up.

Interpretation: confirmatory, i.e. model useful for practice versus exploratory, i.e. more research needed.

Authors’ conclusion

In conclusion, the present study provides a model to individually predict patient’s LR and DM risks during follow-up, applying a flexible parametric competing risk regression approach. These models are at the moment being included in the updated version of the PERSARC app for Individualized Sarcoma Care and follow-up. Although a risk-threshold of 4% for LR and 2% for DM was chosen in the present study, the “optimal” threshold upon which an individual patient should undergo imaging with MRI, chest-CT, or CXR, is still subjected to experts’ opinion and should be further discussed with patients concerned.

Rueten-Budde 2018

Development and internal validation of dynamic PERSARC model

Source of data and date: Clinical data were collected between January 1st, 2000 and December 31st, 2014, at 14 different international specialized sarcoma centers.

Setting/ number of centres and country: Included centers are Leiden University Medical Center (Leiden, the Netherlands), Royal Orthopaedic Hospital (Birmingham and Stanmore, UK), Netherlands Cancer Institute (Amsterdam, the Netherlands), Mount Sinai Hospital (Toronto, Canada), the Norwegian Radium Hospital (Oslo, Norway), Aarhus University Hospital (Aarhus, Denmark), Skane University Hospital (Lund, Sweden), and Medical University Graz (Graz, Austria).

Funding and conflicts of interest:

This work has been supported by the Dutch Cancer Society (DCS) – KWF Kankerbestrijding [UL2015-8028]. The funding source had no role in the design of this study, execution, analyses, interpretation of the data, report writing or decision to submit the article for publication.

Authors Rueten-Budde, van Praag and Fiocco have nothing to disclose. Author van de Sande reports grants from Daiichi Sankyo, outside the submitted work

Recruitment method: consecutive

Inclusion criteria:

Patients were selected from each hospital's own sarcoma registry based on histological diagnosis. Eligible diagnoses included high-grade (FNCLCC grade II and III [11]) angiosarcoma, malignant peripheral nerve sheath tumor (MPNST), synovial sarcoma, spindle cell sarcoma, myxofibrosarcoma, liposarcoma, leiomyosarcoma, malignant fibrous histiocytoma/ undifferentiated pleomorphic sarcoma (MFH/UPS), (pleomorphic) soft tissue sarcomas not-otherwise-specified (NOS), malignant rhabdoid tumor, alveolar soft part sarcoma, epithelioid sarcoma, clear cell sarcoma, rhabdomyosarcoma (adult form) and conventional fibrosarcoma.

Exclusion criteria:

Patients were excluded if they were initially treated without curative intent, presented with LR or DM, had Kaposi's or rhabdomyosarcoma (pediatric form), had a tumor in their abdomen, thorax, head or neck, or received isolated limb perfusion as (neo-) adjuvant treatment.

Treatment:  All patients underwent surgery.

Radiotherapy (%)

• No radiotherapy 916 (41.0)
• Neoadjuvant 265 (11.9)
• Adjuvant 1004 (45.0)
• Unknown 47 ( 2.1)

Chemotherapy (%)

• No chemotherapy 1876 (84.1)
• Neoadjuvant 98 ( 4.4)
• Adjuvant 228 (10.2)
• Unknown 30 ( 1.3)

Participants:

N=2,232

Mean age:

60.86 (SD 18.74)

Sex: % M / % F

53.9/46.1

In the following, baseline and time-dependent variables that were included into the dynamic model are defined. Predictors measured at baseline were: age (years), tumor size by the largest diameter measured at pathological examination (centimeters), tumor depth in relation to investing fascia (deep/superficial), and histological subtype according to WHO classification . Radiotherapy (yes/no) was further specified as being either neoadjuvant or adjuvant treatment. Chemotherapy was not included in the model because it was seldom given to patients for primary tumors. Surgical margins were categorized according to the categorical R-system: ‘R0’ for a negative margin and ‘R1-2’ for a positive margin with tumor cells in the inked surface of the resection margin. The potential effect modifier grade was not included, since all included patients had high-grade tumors. Local recurrence was defined as the presence of pathologically and/or radiologically confirmed tumor at the site where it was originally detected, more than two months after primary surgery. Distant metastases were defined as

radiological evidence of systemic spread of tumor distant from the

primary tumor site.

Development

Modelling method: proportional

landmark supermodel, backward selection procedure

Performance

Calibration measures and 95%CI:

Good model calibration was indicated by a heuristic shrinkage factor equal to 0.996.

Discrimination measures and 95%CI: The discriminative ability of the model was

measured with dynamic cross-validated C-indices of 0.694, 0.777, 0.813, 0.810, 0.798, and 0.781 at 0-, 1-, 2-, 3-, 4-, and 5-years after surgery respectively.

Classification measures:

NR

Evaluation

Method for testing model performance:

Internal validation

Type of outcome: single (dynamic)

Definition and method for measurement of outcome: Dynamic overall survival, defined as time from surgery to death from any cause or last recorded follow-up; dynamic probability of surviving an additional five years from a prediction time point t_p called dynamic overall survival (DOS).

Endpoint or duration of follow-up: until death/NR

Number of events/outcomes:

Median follow-up of 6.42 years (95% confidence interval: 6.17–6.72). In total 1034 patients died, 143 patients developed LR, 556 DM, and 159 developed both.

RESULTS

Multivariable model:

Coefficients: HR (95% CI)

Covariates with time-constant effects

Age (ref: 60 years, per 10 years)

Age 1.444 (1.381–1.510)

Age2 1.065 (1.048–1.082)

Tumor size (ref: 0 cm, per 1 cm)

Size 1.120 (1.072–1.169)

Size2 0.997 (0.996–0.999)

Tumor depth (superficial vs. deep)

0.784 (0.654–0.940)

Radiotherapy (RT)

No RT 1

Neoadjuvant 0.773 (0.572–1.044)

Adjuvant 0.903 (0.763–1.068)

Local recurrence (yes vs. no)

1.998 (1.622–2.461)

Distant metastasis (yes vs. no)

7.572 (6.501–8.818)

Covariates with time-varying effects

Prediction time (ref: time of surgery, per year)

tp 0.431 (0.330–0.562)

tp2 1.127 (1.066–1.192)

Histology Constant

Myxofibrosarcoma 1

MPNST 1.807 (1.270–2.571)

Synovial sarcoma 1.323 (0.971–1.801)

Sarcoma – NOS 1.181 (0.784–1.781)

Spindle cell sarcoma 0.819 (0.638–1.051)

MFH/UPS 1.000 (0.789–1.269)

Other 1.229 (0.828–1.825)

Histology Linear time-varying effect

Myxofibrosarcoma 1

MPNST 0.916 (0.692–1.212)

Synovial sarcoma 1.368 (1.084–1.727)

Sarcoma – NOS 1.067 (0.739–1.540)

Spindle cell sarcoma 1.184 (0.959–1.461)

MFH/UPS 1.256 (1.024–1.540)

Other 1.050 (0.742–1.486)

Histology Quadratic time-varying effect

Myxofibrosarcoma 1

MPNST 0.985 (0.930–1.044)

Synovial sarcoma 0.913 (0.864–0.964)

Sarcoma – NOS 0.983 (0.913–1.058)

Spindle cell sarcoma 0.990 (0.947–1.035)

MFH/UPS 0.968 (0.928–1.010)

Other 0.985 (0.913–1.062)

Margin Constant

R0 vs. R1-2 0.764 (0.606–0.964)

Margin Linear time-varying effect

R0 vs. R1-2 1.417 (1.127–1.783)

Margin Quadratic time-varying effect

R0 vs. R1-2 0.947 (0.902–0.993)

Alternative presentation of final model: The results of this study will be made freely available through the updated PERsonalized SARcoma Care (PERSARC) mobile application.

Interpretation: confirmatory

Authors’ conclusion

The presence of time-varying effects, as well as the effect of local recurrence and distant metastases on survival, suggest the importance of updating predictions during follow-up. This newly developed dynamic prediction model which updates survival probabilities over time can be used to make better individualized treatment decisions based on a dynamic assessment of a patient's prognosis.

Rueten-Budde 2021

Update and external validation of dynamic PERSARC model

Source of data and date:

The model development data were augmented for the update and contained data from Leiden University Medical Center, Royal Orthopaedic Hospital, Netherlands Cancer Institute, Mount Sinai Hospital, the Norwegian Radium Hospital, Aarhus University Hospital, Skåne University Hospital, Medical University Graz, Royal Marsden Hospital, Erasmus MC Cancer Institute, Radboud University Medical Center, University Medical Center Groningen, Haukeland University Hospital, Helios Klinikum Berlin‐Buch, MedUni Vienna, Vienna General Hospital, and the EORTC trial 62931, a randomized controlled trial which studied the effect of intensive adjuvant chemotherapy on several outcome measures. External data were provided by Istituto Nazionale dei Tumori. For both, the model development and external cohort data were collected from centers between January 1, 2000, and December 31, 2014. Data from the EORTC trial 62931, which is part of the development cohort, were collected between February 1995, and December 2003.

Setting/ number of centres and country: see above

Funding and conflicts of interest:

ACKNOWLEDGMENT

This study has been supported by the Dutch Cancer Society (DCS) – KWF Kankerbestrijding (Grant no. UL2015‐8028). The funding source had no

role in the design of this study, execution, analyses, interpretation of the data, report writing, or decision to submit the article for publication.

CONFLICT OF INTERESTS

Authors Anja J. Rueten‐Budde, Veroniek M. van Praag, and Marta Fiocco have nothing to disclose. Author Michiel A. J. van de Sande reports grants from Daiichi Sankyo, outside the submitted work.

Recruitment method: consecutive

Inclusion criteria:

Selection and exclusion criteria were identical for the model development (update) cohort and the external cohort.

Included eSTS subtypes included high‐grade (FNCLCC Grades II and III) angiosarcoma, malignant peripheral nerve sheath tumor, synovial sarcoma, spindle cellsarcoma, myxofibrosarcoma, liposarcoma, leiomyosarcoma, malignant fibrous histiocytoma/ undifferentiated pleomorphic sarcoma, (pleomorphic) soft tissue sarcomas not‐otherwise‐specified, epithelioid sarcoma, clear cell sarcoma, rhabdomyosarcoma (adult form), conventional fibrosarcoma, giant cell sarcoma, malignant granular cell tumor, unclassified soft tissue sarcoma, and undifferentiated sarcoma.

Exclusion criteria:

Patients were excluded if they were initially treated without curative

intent, presented with LR or DM, had Kaposi's or rhabdomyosarcoma (pediatric form), had tumor in their abdomen, thorax, head, or neck, or

received isolated limp perfusion as (neo‐) adjuvant treatment.

Treatment: All patients underwent resection.

Radiotherapy (%)

No radiotherapy

UC: 1331 (34.8)

VC: 474 (42.7)

Neoadjuvant

UC: 517 (13.5)

VC: 138 (12.4)

Adjuvant

UC: 1878 (49.1)

VC: 499 (44.9)

Unknown

UC: 100 (2.6)

VC: 0 (0.0)

Chemotherapy (%)

No

UC: 3189 (83.4)

VC: 739 (66.5)

Yes

UC: 470 (12.3)

VC: 372 (33.5)

Unknown

UC: 167 (4.4)

VC: 0 (0.0)

Participants:

Update cohort (UC) N=3,826

Validation cohort (VC) N=1,111

Mean age (SD):

UC: 59.40 (18.10)

VC: 55.46 (17.03)

Sex: % M / % F

UC: 52.6/43.9 (3.5 unknown)

VC: 54.6/45.4

The dynamic prediction model developed in Rueten‐Budde (2018)

was revised by adding more patients and the variable grade to the

model.

Development

Modelling method: N/A

Performance

Calibration measures and 95%CI:

VC: calibration plot, author concluded that the figure shows they are relatively close to the diagonal line implying that predictions are accurate; the model generally slightly underestimated survival.

Discrimination measures and 95%CI:

VC: The discriminative ability of the model was assessed with dynamic C‐indices, with values equal to 0.697, 0.790, 0.822, 0.818, 0.812, and 0.827 at 0, 1, 2, 3, 4, and 5 years after surgery respectively.

Classification measures:

NR

Evaluation

Method for testing model performance:

External validation

Type of outcome: single (dynamic)

Definition and method for measurement of outcome:

The outcome of interest was OS, defined as the time from surgery to death due to any cause or last recorded follow‐up. The dynamic model predicts 5‐year dynamic overall survival (DOS) from a

particular prediction time point during follow‐up.

Endpoint or duration of follow-up: until death/NR

Number of events/outcomes:

UC: median follow‐up equal to 6.00 years (95% confidence interval [CI] = 5.86–6.18)

VC: median follow‐up equal to 6.89 years (95% CI = 6.47–7.61).

In the development cohort (update), in total 1602 patients died, 241 patients developed LR, 949 DM, and

385 developed both. In the external cohort, 306 patients died, 70 had LR, 279 DM, and 77 developed both.

RESULTS

Multivariable model:

Revised model reported.

Alternative presentation of final model: The updated dynamic prediction models is implemented in the

updated PERSARC application; available for free at the Apple Store and Google Play Store.

Interpretation: confirmatory

Authors’ conclusion

Results from the external validation show that the dynamic PERSARC model is reliable in predicting the probability of surviving an additional 5 years from a specific prediction time point during follow‐up. The model combines patient‐, treatment‐specific and time‐dependent variables such as local recurrence and distant metastasis to provide accurate survival predictions throughout follow‐up and is available through the PERSARC app.

Study reference

(first author, year of publication)

Classification¹

Participant selection

1) Appropriate data sources?²

2) Appropriate in- and exclusion?

Risk of bias: low/high/unclear

Predictors

1) Assessed similar for all participants?

2) Assessed without knowledge of outcome?

3) Available at time the model is intended to be used?

Risk of bias: low/high/unclear

Outcome

1) Pre-specified or standard outcome definition?

2) Predictors excluded from definition?

3) Assessed similar for all participants?

4) Assessed without knowledge of predictors?

5) Time interval between predictor and outcome measurement appropriate?

Risk of bias: low/high/unclear

Analysis

1) Reasonable number of participants with event/outcome?

2) All enrolled participants included in analysis?

3) Missing data handled appropriately?

4) No selection of predictors based on univariate analysis?

5) Relevant model performance measures evaluated appropriately?³

6) Accounted for model overfitting⁴ and optimism?

7) Predictors and weights correspond to results from multivariate analysis?

Risk of bias: low/high/unclear

Overall judgment

High risk of bias: at least one domain judged to be at high risk of bias.

Model development only: high risk of bias.

Risk of bias: low/high/unclear

MSKCC; Kattan, 2002; Eilber, 2004; Mariani, 2005; Squires 2022; development and external validation of model

Low

(Data obtained from databases in which patients were prospectively entered, consecutive patients or all patients who underwent resection of primary extremity STS in different centers during time period. Clear in- and exclusion criteria)

Unclear, probably high

(Clear definitions of predictors. Patients from multiple centers, predictors may have been recorded differently at different centers. Predictors recorded before the outcome occurred.)

Unclear, probably low

(Outcome is sarcoma-specific death, may be misclassified. No information on whether assessor of outcome was aware of predictors.)

High

(Patients with missing values were excluded (n=139); and for other studies no information on missing data. No effect sizes reported for the predictors in the developed nomogram.)

High risk of bias

SAM; Sampo 2012; development and external validation of model

Unclear, probably low

(For development: all patients referred to STS Sarcoma Group in time period, data probably obtained from register but not explicitly described. Data for validation cohort obtained from hospital database. Both with clear in- and exclusion criteria)

Unclear, probably low

(Predictors are clearly defined and assessed in the same way for all study participants. Predictors were recorded before the outcome occurred. Not explicitly reported whether re-evaluation/re-assessment was blinded.)

Unclear, probably low

(Outcome sarcoma-specific survival, may be misclassified. No information on whether assessor of outcome was aware of predictors.)

High

(Missing data for 84 patients in development cohort, patients excluded. Validation cohort: 224 patients excluded, unclear how many due to missing data; “patients with metastatic disease at presentation, patients receiving adjuvant chemotherapy and patients with missing data on the assessed and reported parameters were excluded”)

High risk of bias

Sarculator; Callegaro, 2016; Callegaro, 2019; Squires, 2022; Voss, 2022; development and external validation of model

Low

(Data obtained from institutional or national prospectively maintained databases.

Clear in- and exclusion criteria)

Unclear

(Predictors are clearly defined and assessed in the same way for all study participants. Predictors recorded before outcome.

Squires: Patients from multiple centers, predictors may have been recorded differently at different centers.

Voss: data from national database; predictors may have been recorded differently at different centers.)

Low

(Outcome is overall survival, not likely to be misclassified. No information on whether assessor of outcome was aware of predictors.)

Unclear

(No information on missing data for Callegaro 2016 and Squires.

Callegaro 2019; 12 patients excluded because survival time was missing, small percentage of the total of 3,740.

Voss: patients excluded with incomplete grade, treatment, or survival data, not mentioned how many.)

Some concerns

PERSARC; Van Praag, 2017; Smolle, 2019; Rueten-Budde, 2018; Rueten-Budde, 2021;  development and external validation of model

Low

(Data obtained from prospective sarcoma databases.

Clear in- and exclusion criteria)

Unclear

Low

(Outcome overall survival is not likely to be misclassified. Outcome local recurrence, clear definition, misclassification not likely. No information on whether assessor of outcome was aware of predictors.)

Unclear

Van Praag: Due to missing values for 72 patients, 766 individuals were included

Some concerns