Original article

Multicentre study of the learning curve and surgical performance of cytoreductive surgery with intraperitoneal chemotherapy for pseudomyxoma peritonei S. Kusamura1 , B. J. Moran2 , P. H. Sugarbaker3 , E. A. Levine4 , D. Elias6,7 , D. Baratti1 , D. L. Morris9 , A. Sardi5 , O. Glehen7,8 and M. Deraco1 , on behalf of Peritoneal Surface Oncology Group International (PSOGI)* 1

Peritoneal Surface Malignancy Programme, Colorectal Cancer Unit, IRCCS Fondazione Istituto Nazionale Tumori di Milano, Milan, Italy, Basingstoke and North Hampshire NHS Foundation Trust, Basingstoke, UK, 3 Washington Cancer Institute, Washington Hospital Center, Washington, DC, 4 Surgical Oncology Service, Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, and 5 Division of Surgery, Department of Surgical Oncology, Institute for Cancer Care, Mercy Medical Center, Baltimore, Maryland, USA, 6 Department of Surgical Oncology, Institut Gustave Roussy, Cancer Centre, Villejuif, and 7 RENAPE: Centre Expert National de Référence des Cancers Rares du Péritoine, Unité de Recherche Clinique, Centre Hospitalier Universitaire (CHU) de Lyon Sud, and 8 Department of Digestive Surgery, CHU de Lyon Sud, Hospices Civils de Lyon, Pierre Bénite, France, and 9 Hepatobiliary and Surgical Oncology Unit, University of New South Wales Department of Surgery, St George Hospital, New South Wales, Sydney, Australia Correspondence to: Dr M. Deraco, IRCCS Fondazione Istituto Nazionale dei Tumori di Milano, via Venezian 1, 20133 Milan, Italy (e-mail: [email protected]) 2

Background: The learning curves for cytoreductive surgery with intraperitoneal chemotherapy for

treatment of pseudomyxoma peritonei (PMP) were explored between international centres/surgeons to identify institutional or other factors that might affect performance. Methods: Data from patients with PMP treated with the combined procedure across 33 international centres between 1993 and 2012 were analysed retrospectively. A risk-adjusted sequential probability ratio test was conducted after defining the target outcome as early oncological failure (disease progression within 2 years of treatment), an acceptable risk for the target outcome (odds ratio) of 2, and type I/II error rates of 5 per cent. The risk prediction model was elaborated and patients were evaluated sequentially for each centre/surgeon. The learning curve was considered to be overcome and proficiency achieved when the odds ratio for early oncological failure became smaller than 2. Results: Rates of optimal cytoreduction, severe postoperative morbidity and early oncological failure were 84⋅4, 25⋅7 and 29⋅0 per cent respectively. The median annual centre volume was 17 (range 6–66) peritoneal malignancies. Only eight of the 33 centres and six of 47 surgeons achieved proficiency after a median of 100 (range 78–284) and 96 (86–284) procedures respectively. The most important institutional factor affecting surgical performance was centre volume. Conclusion: The learning curve is extremely long, so centralization and/or networking of centres is necessary to assure quality of services. One centre for every 10–15 million inhabitants would be ideal. ∗ Members of PSOGI are co-authors of this study and can be found under the heading Collaborators Presented to the 66th Society of Surgical Oncology Annual Cancer Symposium, National Harbor, Maryland, USA, February 2013

Paper accepted 17 September 2014 Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9674

Introduction

Pseudomyxoma peritonei (PMP) is a rare neoplastic process (approximately 1–2 patients per million per year) characterized by disseminated peritoneal mucinous tumour deposition with progressive mucinous ascites1,2 . Since its advent in the early 1990s, the combination of © 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

cytoreductive surgery with intraperitoneal chemotherapy has been used widely for the treatment of peritoneal surface malignancies, including PMP3,4 . Surgical performance for cytoreductive surgery plus intraperitoneal chemotherapy was monitored recently in a series of studies carried out at two Italian centres (National BJS

S. Kusamura, B. J. Moran, P. H. Sugarbaker, E. A. Levine, D. Elias, D. Baratti et al.

Cancer Institute in Milan and Bentivoglio Hospital) using the risk-adjusted sequential probability ratio test5 – 7 . Setting ability to achieve a complete cytoreduction, morbidity and mortality as target outcomes, it was shown that approximately 140 procedures are necessary to attain surgical expertise. Moreover, it was concluded that the type of surgical training might shorten the length of the learning process. These analyses were based on experiences at only two centres and comprised several histological types of peritoneal carcinomatosis. The aim of the present analysis was to test the validity and reproducibility of previous results in other international centres, comparing the learning curve for cytoreductive surgery with intraperitoneal chemotherapy between surgeons and institutions. Rather than morbidity, mortality or completeness of cytoreduction, surgical performance was evaluated by means of a prognostic outcome. The second aim was to evaluate factors related to each institution and the surgeon’s background, in addition to biological tumour characteristics and the nature of the treatment, which could affect prognosis and surgical performance. Methods

This retrospective study was conducted by the Peritoneal Surface Oncology Group International (PSOGI) using a multicentre registry of patients with PMP who were treated between February 1993 and February 2012. Thirty-three peritoneal surface malignancy centres from around the world and 47 surgeons were included in this evaluation. The inclusion criterion for the study was histologically confirmed peritoneal dissemination from an appendiceal mucinous neoplasm. The exclusion criteria included the presence of colorectal malignancies and/or extra-abdominal metastases.

Cytoreductive surgery with intraperitoneal chemotherapy (hyperthermic and/or early postoperative intraperitoneal chemotherapy) Cytoreductive surgery involves a combination of peritonectomy procedures and visceral resections performed with the aim of resecting all intra-abdominal tumour implants with a diameter greater than 2⋅5 mm1 . Hyperthermic intraperitoneal chemotherapy (HIPEC) was administered following cytoreduction using an open coliseum or closed technique, with the perfusate heated to between 40 and 42∘ C. Some centres also administered early postoperative intraperitoneal chemotherapy comprising 5-flurouracil (650 mg/m2 ) on days 1–5 after surgery. © 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

Study parameters The outcome of interest was early oncological failure, defined as disease progression within 2 years of surgery. Progression was defined as relapse after complete cytoreduction, progressive disease after incomplete cytoreduction, or death from any cause during follow-up. Death from any cause was adopted instead of cause-specific mortality. The latter, although of major biological importance in a disease-specific intervention, is a more subjective endpoint than total mortality, and may also miss important effects of therapy that could actually shorten overall survival. For example, death from postoperative myocardial infarction, although indirectly related to the treatment, would not change the disease-specific mortality8 . The variables of interest were age, sex, Eastern Cooperative Oncology Group performance status, previous systemic chemotherapy, and previous surgical score measuring the magnitude of surgical manipulation before the combined procedure (0, no surgery or biopsy; 1, surgery in 1 abdominal region only; 2, surgery in 2–5 regions; 3, surgery in more than 5 regions)9 . Variables related to disease characteristics were histological subtype (disseminated peritoneal adenomucinosis, peritoneal mucinous carcinomatosis (PMCA), and PMCA with intermediate features)10 , as well as disease extension measured by the Peritoneal Cancer Index9 . Surgery-related variables included completeness of cytoreduction, grade 3–5 morbidity according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 311 , HIPEC and early postoperative intraperitoneal chemotherapy. Procedure-related mortality was defined as death occurring during the hospital stay after cytoreductive surgery with intraperitoneal chemotherapy. Parameters related to the surgeons’ backgrounds included the type of training received during residency (surgical oncology, gastrointestinal surgery, general surgery, other surgical specialties) and the type of fellowship training attended to learn the combined procedure (no fellowship, external institutional fellowship, local institutional fellowship). Only the pioneers who developed this combined treatment were classified as having not attended a fellowship guided by a more experienced surgeon. Knowledge transfer occurred in an external institutional fellowship when the surgeon learned the technique as a visiting doctor in centres that were already specialized. When a surgeon who was already proficient taught the technique, the knowledge transfer was classified as a local institutional fellowship. Variables related to the centres were: number of principal surgeons participating in the peritoneal surface malignancy programme; centre volume, defined as the annual caseload www.bjs.co.uk

BJS

Surgical performance and learning curve of cytoreductive surgery

Clinicopathological characteristics of patients with pseudomyxoma peritonei undergoing cytoreductive surgery and intraperitoneal chemotherapy

0

Age (years)† Sex ratio (M : F) Previous surgical score 0 1 2 3 Previous systemic chemotherapy Yes No Tumour histological subtype DPAM PMCA I/D PMCA Peritoneal Cancer Index†

No. of patients* (n = 2451)

Missing data (%)

53⋅1(12⋅0) 1032 : 1419

16⋅6 0 22⋅4

403 (16⋅4) 444 (18⋅1) 580 (23⋅7) 476 (19⋅4) 39⋅1

–0·5

–1·0

–1·5

–2·0

395 (16⋅1) 1098 (44⋅8) 3⋅6 1486 (60⋅6) 191 (7⋅8) 685 (27⋅9) 20(1)

0

10

20

30

40

50

60

70

Centre volume

40⋅0

*With percentages in parentheses unless indicated otherwise; †values are mean(s.d.). DPAM, disseminated peritoneal adenomucinosis; PMCA I/D, peritoneal mucinous carcinomatosis with intermediate features; PMCA, peritoneal mucinous carcinomatosis.

of peritoneal malignancies (not only PMP) treated with cytoreductive surgery plus intraperitoneal chemotherapy; and the proportion of patients with PMP treated at each centre relative to the total number of peritoneal surface malignancies treated there during the study interval.

Statistical analysis Assessment of the effects of oncological outcomes on the learning curve and surgical performance required the adoption of a discrete surrogate marker of survival (early oncological failure)12 . The missing data were handled using the multiple imputation technique and the multivariable imputation with chained equations (MICE) approach13 – 17 . This is a powerful statistical tool that generates plausible data based on the distribution of other variables from the data set and minimizes the loss of power owing to missing information. Correlations between continuous predictors and the outcome variable were studied with restricted cubic spline functions18 . Next, a multivariable logistic regression model with backward elimination of predictors was carried out19 . The final (pooled) model was fitted on the basis of multiply imputed data sets using Rubin’s rules to combine estimates of effects and their standard errors20 . The quality of the multivariable model was assessed by calculating its calibration and discrimination21,22 . The sequential probability ratio test has been used recently in healthcare contexts to monitor the performance of medical and surgical interventions. In contrast © 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

Log odds of early oncological failure

Table 1

Centre volume (annual caseload of peritoneal malignancies) versus log-transformed risk of early oncological failure among 2451 patients with pseudomyxoma peritonei treated by cytoreductive surgery and intraperitoneal chemotherapy. The non-linear correlation between these variables was evaluated using restricted cubic splines with five knots

Fig. 1

to traditional frequentist statistics23 , this method allows sequential data analysis within a framework of hypothesis testing, can be applied to analyse a learning curve, and allows risk adjustment24 – 28 . Comparing unadjusted event rates for different centres or surgeons would unfairly penalize those performing operations on higher-risk patients. Therefore, a risk adjustment was conducted to render the comparison of surgical performance between centres and surgeons feasible. Risk adjustment is a corrective tool that adjusts for the differences in risk among specific patients in order to level the playing field regarding the reporting of patient outcomes. The predicted risks for each patient from the imputed data sets were averaged and applied to the log likelihood ratio calculation of risk-adjusted sequential probability ratio test curves. The parameters were defined as follows: type I and II error rates = 0⋅05 and unacceptable odds ratio (OR) for early oncological failure = 2. From these, two control limits (‘accept’ and ‘reject’ lines) and the risk-adjusted sequential probability ratio test curves were plotted, according to the equations listed in the Appendix S1 (supporting information)27,29 . The key point for interpretation of the risk-adjusted sequential probability ratio test is whether the curve breaches the boundary lines. When the curve crosses the upper decision limit (‘reject line’) from below, the alternative hypothesis is accepted, meaning that the actual risk of early oncological failure is greater than 2. On the other hand, if the line crosses the lower decision limit (‘accept line’) from above, the null hypothesis is accepted, which implies that: the actual risk of early www.bjs.co.uk

BJS

Cumulative log likelihood ratio

S. Kusamura, B. J. Moran, P. H. Sugarbaker, E. A. Levine, D. Elias, D. Baratti et al.

Learning phase

Audit phase

10·0 h1 5·0 0·0 –5·0

Breaking point of learning curve: 81st patient

h0 –10·0

0

40

80

120

160

200

240

280

No. of operations

Centre 4 Cumulative log likelihood ratio

a

10·0 h1 Poor performance with unacceptable risk of early oncological failure

5·0 0·0 –5·0

Breaking point of learning curve: 116th patient

h0 –10·0

0

20

40

60

80

100

120

140

No. of operations

Cumulative log likelihood ratio

b

Centre 9 10·0 h1

5·0 0·0 Proficiency not achieved

–5·0

h0 –10·0 0

10

20

30

40

No. of operations

c

Centre 1

Fig. 2 Examples of sequential probability ratio test control charts for a centre 4, b centre 9 and c centre 1. The target outcome for monitoring surgical performance was early oncological failure, defined as disease progression within 2 years of operation. Patients with less than 24 months of follow-up were excluded. h0, Lower boundary; h1, upper boundary. A curve with a descending feature means improvement in surgical performance whereas an ascending feature indicates deterioration. The breaking point of the learning curve is the point at which the curve crosses h0. When the curve surpasses any boundary line (h0 or h1) it is reset to 0 to restart the surveillance of surgical performance

oncological failure is not greater than 2; the surgical team has overcome the learning curve; and this point is the breaking point of the learning curve. After the breaking point, the curve is reset to zero to restart the monitoring with intent to audit. The same risk prediction model was used to plot curves for centres and surgeons. Monitoring surgical performance should continue even after expertise has been acquired, as eventual deterioration could emerge owing to several factors, such as the introduction of new surgeons to the centre’s team or the adoption of new treatment protocols. The statistical analyses for multiple imputations and risk prediction modelling were conducted © 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

using R software version 3.0.1 (http://www.r-project.org), and the risk-adjusted sequential probability ratio test was done in Microsoft® Excel version 2003 (Microsoft Corporation, Redmond, Washington, USA). Statistical significance was set at P < 0⋅050. Further details of all statistical analyses can be found in Appendix S1 (supporting information). Results

The demographic and clinicopathological data of the 2451 patients analysed are summarized in Table 1. Regarding the www.bjs.co.uk

BJS

Surgical performance and learning curve of cytoreductive surgery

short-term outcomes, 2069 patients (84⋅4 per cent) had CC-0/1 or optimal cytoreduction after surgery. Some 2231 patients (91⋅0 per cent) underwent HIPEC. The mean(s.d.) amount of red cells transfused during surgery was 3⋅2(6⋅8) units. The median (i.q.r.) duration of intensive care unit and hospital stay was 2 (1–5) and 16 (11–24) days respectively. A total of 702 patients (28⋅6 per cent) underwent early postoperative intraperitoneal chemotherapy. Rates of grade 3–5 morbidity and early oncological failure were 25⋅7 and 29⋅0 per cent respectively. The centres had a median volume (annual caseload) of 17 (range 6–66) patients (Table S1, supporting information). The median proportion of patients with PMP among peritoneal surface malignancy cases across centres was 0⋅32 (range 0⋅03–0⋅69). Of the 47 surgeons who participated in this study, 14 identified themselves as surgical oncologists, 13 were general surgeons, 11 were gastrointestinal surgeons, one was a gynaecologist, and eight did not state their training specialization. Three surgeons declared that they did not attend any type of fellowship to learn how to perform cytoreductive surgery with intraperitoneal chemotherapy. Nineteen were trained on the basis of an external institutional fellowship and 12 underwent a local institutional fellowship. Twenty-eight centres had only one surgeon, whereas five centres had two or more surgeons doing cytoreductive surgery with intraperitoneal chemotherapy. The pooled risk prediction model identified the following independent predictors of early oncological failure: centre volume, proportion of PMP, number of principal surgeons on the team, site of fellowship, previous systemic chemotherapy, histological subtype, Peritoneal Cancer Index, completeness of cytoreduction, HIPEC and early postoperative intraperitoneal chemotherapy (Table S2, supporting information). The volume–outcome correlation was significant and non-linear, and the restricted cubic spline transformation was applied to identify this relationship. Patients were plotted on a graph of centre volume versus OR for early oncological failure (with logarithmic transformation). The line representing the link assumed a curvilinear feature with a wide range of oscillations until the point at which centre volume increased above 59, when the curve started to descend sharply, reflecting a consistent and significant reduction in the risk of early oncological failure (Fig. 1). Only eight of the 33 centres overcame the learning curve, and after a median of 100 (range 78–284) procedures (Table S1, supporting information). Only six of the 47 surgeons in these centres overcame the learning curve, after a median of 96 (range 86–284) procedures. Examples of risk-adjusted sequential probability ratio tests are shown

in Fig. 2. After gaining expertise, no centre or surgeon was reported as having an unacceptable deterioration in surgical performance.

© 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

Discussion

The prognostic aspect related to surgical performance was assessed in this study, as prognosis has been deemed the ultimate testament to a learning process30 . Early oncological failure was chosen as the outcome of interest because it represents not only a surrogate marker of oncological outcome31 , but also a single parameter that encompasses other aspects included in the concept of expertise (such as capacity to choose the right candidate for the combined treatment by means of a judicious preoperative evaluation32 , to cytoreduce PMP completely in a safe way, and to conduct postoperative care in a multidisciplinary environment). The authors did not focus on postoperative morbidity to evaluate surgical performance, as prognosis has been shown to be a function of an uneventful postoperative period3 . Therefore, choosing early oncological failure as the target outcome provided an indirect evaluation of surgical performance in terms of safety. The learning phases in the different centres turned out to be extremely long. Considering the high morbidity rates33 and treatment costs34 , developing strategies to shorten the learning curve is critically important. Shortening the learning curve would imply a reduction in the number of complications resulting from inexperience, as well as a reduction in costs associated with procedures performed by non-experts. A number of factors might have accounted for such a lengthy learning process. First, the procedure is demanding and complex. Second, most surgeons in this study who achieved proficiency attended an external observer-based fellowship, and this type of knowledge transfer is less effective than an operative local institutional fellowship6 . Third, the learning phase of expert centres or surgeons overlapped with a historical period (1993–2004) in which the procedure was still evolving conceptually. In fact, the article35 establishing that a right colonic resection during the cytoreduction of appendiceal carcinoma does not confer a survival advantage was published in 2004. Finally, a factor that might have influenced the length of the learning process, owing to its independent correlation with early oncological failure, was the number of principal operators per centre. Centres with more than three principal surgeons on the team were at higher risk of their performance worsening over time, probably because of the protracting effect on the local institutional knowledge transfer. Some argue against the need for centralization, claiming that the combined procedure is technically quite similar BJS

S. Kusamura, B. J. Moran, P. H. Sugarbaker, E. A. Levine, D. Elias, D. Baratti et al.

when applied to other kinds of peritoneal malignancy. To test whether peritoneal surgeons need specific training for PMP, the effect of proportion of PMPs in the centre’s experience on prognosis was evaluated. This revealed that the greater the institution’s specialization in treating PMP, the lower the rate of early oncological failure. This finding is in line with the situation observed in the UK, where peritoneal surface malignancy programmes are specialized in PMP and located in only two referral centres that provide services for the entire country36 . Another independent predictor of target outcome was centre volume. A non-linear association between centre volume and early oncological failure was identified, and those with more than 59 procedures had the lowest risk. This finding is in apparent conflict with a previous analysis of this data set that found no difference in survival outcome between ‘well established’ and ‘emerging’ centres treating PMP with cytoreductive surgery and intraperitoneal chemotherapy3 . This paradox is not surprising as most centres deemed to be well established had only recently overcome the learning curve, according to the present analysis. This study has some limitations. Data were retrospective and in some parts incomplete. The latter drawback was partly solved by using the multiple imputation approach, but a number of important parameters with potential prognostic impact (such as co-morbidities, preoperative performance status and baseline serum tumour markers) were not included in the analysis, as they were not in the original data set. Nevertheless, it is possible to conclude that at least 100 procedures per centre and 96 per surgeon are necessary for acquiring optimal expertise in the management of PMP. These figures are possible only for centres serving large populations. Considering that the approximate annual incidence of PMP is 1–2 million, it can be estimated reasonably that one centre for every 10–15 million inhabitants would be ideal. An alternative to centralization as a means of improving the quality of services is to stimulate mutual cooperation of centres to favour the referral of patients with PMP to institutions with more experience. In fact, most of the low-volume centres in this study are affiliated to RENAPE, the French network for the treatment of rare peritoneal diseases37 . Collaborators

Other members of the PSOGI: F. N. Gilly (Department of Digestive Surgery, Centre Hospitalier Universitaire (CHU) de Lyon Sud, Hospices Civils de Lyon, and RENAPE, Unité de Recherche Clinique, CHU de Lyon Sud, Pierre-Benite, France), P. Barrios (Department of Oncological Surgery, Hospital Sant Joan Despí, © 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

Moises Broggi, Peritoneal Surface Malignancy Catalonian’s Programme, Sant Joan Despí, Barcelona, Spain), F. Quenet (Centre Régional de Lutte du Cancer Val d’Aurell, Montpellier, and RENAPE, CHU de Lyon Sud, Pierre-Benite, France), B. W. Loggie (Division of Surgical Oncology, Creighton University Medical Center, Omaha, New England, USA), A. Gómez Portilla (Department of General Surgery and Digestive Diseases, Hospital Santiago Apostol, Vitoria, Spain), I. H. J. T. de Hingh (Department of Surgery, Catharina Hospital, Eindhoven, The Netherlands), W. P. Ceelen (Department of Gastrointestinal Surgery, University Hospital, Ghent, Belgium), J. O. W. Pelz (Department of General, Visceral and Paediatric Surgery, University of Wuerzburg, Wuerzburg, Germany), P. Piso (Department of Surgery, University Medical Centre Regensburg, Regensburg, Germany), S. González-Moreno (Peritoneal Surface Oncology Programme, Department of Surgical Oncology, M. D. Anderson Cancer Center Madrid, Madrid, Spain), K. Van Der Speeten (Department of Surgical Oncology, Ziekenhuis Oost-Limburg, Genk, Belgium), T. C. Chua (Hepatobiliary and Surgical Oncology Unit, University of New South Wales Department of Surgery, St George Hospital, Sydney, New South Wales, Australia), T. D. Yan (Department of Cardiothoracic Surgery, University of Sydney, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia) and W. Liauw (Hepatobiliary and Surgical Oncology Unit, University of New South Wales Department of Surgery, St George Hospital, Sydney, New South Wales, Australia) Acknowledgements

The authors acknowledge the contributions of RENAPE, I. Bonnefoy (Centre Hospitalo-Universitaire Lyon Sud, Pierre-Benite, France), M. Cromer (Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, USA) and K. Chandrakumaran (Basingstoke and North Hampshire NHS Foundation Trust, Basingstoke, UK). References 1 Sugarbaker PH. New standard of care for appendiceal epithelial neoplasms and pseudomyxoma peritonei syndrome? Lancet Oncol 2006; 7: 69–76. 2 Ronnett BM, Shmookler BM, Sugarbaker PH, Kurman RJ. Pseudomyxoma peritonei: new concepts in diagnosis, origin, nomenclature, and relationship to mucinous borderline (low malignant potential) tumors of the ovary. Anat Pathol 1997; 2: 197–226. 3 Chua TC, Moran BJ, Sugarbaker PH, Levine EA, Glehen O, Gilly FN et al. Early- and long-term outcome data of

www.bjs.co.uk

BJS

Surgical performance and learning curve of cytoreductive surgery

4

5

6

7

8

9

10

11

12

13

14

15

16

patients with pseudomyxoma peritonei from appendiceal origin treated by a strategy of cytoreductive surgery and hyperthermic intraperitoneal chemotherapy. J Clin Oncol 2012; 30: 2449–2456. McDonald JR, O’Dwyer ST, Rout S, Chakrabarty B, Sikand K, Fulford PE et al. Classification of and cytoreductive surgery for low-grade appendiceal mucinous neoplasms. Br J Surg 2012; 99: 987–992. Kusamura S, Baratti D, Deraco M. Multidimensional analysis of the learning curve for cytoreductive surgery and hyperthermic intraperitoneal chemotherapy in peritoneal surface malignancies. Ann Surg 2012; 255: 348–356. Kusamura S, Baratti D, Virzì S, Bonomi S, Iusco DR, Grassi A et al. Learning curve for cytoreductive surgery and hyperthermic intraperitoneal chemotherapy in peritoneal surface malignancies: analysis of two centres. J Surg Oncol 2013; 107: 312–319. Kusamura S, Baratti D, Hutanu I, Rossi P, Deraco M. The importance of the learning curve and surveillance of surgical performance in peritoneal surface malignancy programs. Surg Oncol Clin N Am 2012; 21: 559–576. National Cancer Institute. Levels of Evidence for Adult and Pediatric Cancer Treatment Studies (PDQ®). http://www. cancer.gov/cancertopics/pdq/levels-evidence-adulttreatment/HealthProfessional/page3 [accessed 1 December 2013]. Jacquet P, Sugarbaker PH. Current methodologies for clinical assessment of patients with peritoneal carcinomatosis. J Exp Clin Cancer Res 1996; 15: 49–58. Ronnett BM, Zahn CM, Kurman RJ, Kass ME, Sugarbaker PH, Shmookler BM. Disseminated peritoneal adenomucinosis and peritoneal mucinous carcinomatosis. A clinicopathologic analysis of 109 cases with emphasis on distinguishing pathologic features, site of origin, prognosis, and relationship to ‘pseudomyxoma peritonei’. Am J Surg Pathol 1995; 19: 1390–1408. Younan R, Kusamura S, Baratti D, Cloutier AS, Deraco M. Morbidity, toxicity, and mortality classification systems in the local regional treatment of peritoneal surface malignancy. J Surg Oncol 2008; 98: 253–257. Biswas P, Kalbfleisch JD. A risk-adjusted CUSUM in continuous time based on the Cox model. Stat Med 2008; 27: 3382–3406. Steyerberg EW, van Veen M. Imputation is beneficial for handling missing data in predictive models. J Epidemiol Community Health 2007; 60: 979. Clark TG, Altman DG. Developing a prognostic model in the presence of missing data: an ovarian cancer case study. J Epidemiol Community Health 2003; 56: 28–37. Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? Int J Methods Psychiatr Res 2011; 20: 40–49. Vergouwe Y, Royston P, Moons KG, Altman DG. Development and validation of a prediction model with missing predictor data: a practical approach. J Clin Epidemiol 2010; 63: 205–214.

© 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

17 van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by chained equations in R. J Stat Software 2011; 45: 1–67. 18 Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989; 8: 551–561. 19 Atkinson AC. A note on the generalized information criterion for choice of a model. Biometrika 1980; 67: 413–418. 20 Royston P. Multiple imputation of missing values. Stata J 2004; 4: 227–241. 21 Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: 29–36. 22 Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982; 115: 92–106. 23 Smeenk RM, Verwaal VJ, Zoetmulder FA. Learning curve of combined modality treatment in peritoneal surface disease. Br J Surg 2007; 94:1408–1414. 24 Matheny ME, Ohno-Machado L, Resnic FS. Risk-adjusted sequential probability ratio test control chart methods for monitoring operator and institutional mortality rates in interventional cardiology. Am Heart J 2008; 155: 114–120. 25 Axelrod DA, Guidinger MK, Metzger RA, Wiesner RH, Webb RL, Merion RM. Transplant quality assessment using a continuously updatable, risk-adjusted technique (CUSUM). Am J Transplant 2006; 6: 313–623. 26 Smith IR, Rivers JT, Mengersen KL, Cameron J. Performance monitoring in interventional cardiology: application of statistical process control to a single-site database. EuroIntervention 2011; 6: 955–962. 27 Rogers CA, Reeves BC, Caputo M, Ganesh JS, Bonser RS, Angelini GD et al. Control chart methods for monitoring cardiac surgical performance and their interpretation. J Thorac Cardiovasc Surg 2004; 128: 811–819. 28 Caputo M, Reeves BC, Rogers CA, Ascione R, Angelini GD. Monitoring the performance of residents during training in off-pump coronary surgery. J Thorac Cardiovasc Surg 2004; 128: 907–915. 29 Spiegelhalter D, Grigg O, Kinsman R, Treasure T. Risk-adjusted sequential probability ratio tests: applications to Bristol, Shipman and adult cardiac surgery. Int J Qual Health Care 2003; 15: 7-13. 30 Moradi BN III, Esquivel J. Learning curve in cytoreductive surgery and hyperthermic intraperitoneal chemotherapy. J Surg Oncol 2009; 100: 293–296. 31 Chua TC, Liauw W, Morris DL. Early recurrence of pseudomyxoma peritonei following treatment failure of cytoreductive surgery and perioperative intraperitoneal chemotherapy is indicative of a poor survival outcome. Int J Colorectal Dis 2012; 27: 381–389. 32 Iversen LH, Rasmussen PC, Laurberg S. Value of laparoscopy before cytoreductive surgery and hyperthermic intraperitoneal chemotherapy for peritoneal carcinomatosis. Br J Surg 2013; 100: 285–292.

www.bjs.co.uk

BJS

S. Kusamura, B. J. Moran, P. H. Sugarbaker, E. A. Levine, D. Elias, D. Baratti et al.

33 Elias D, Honoré C, Ciuchendéa R, Billard V, Raynard B, Lo Dico R et al. Peritoneal pseudomyxoma: results of a systematic policy of complete cytoreductive surgery and hyperthermic intraperitoneal chemotherapy. Br J Surg 2008; 95: 1164–1171. 34 Baratti D, Scivales A, Balestra MR, Ponzi P, Di Stasi F, Kusamura S et al. Cost analysis of the combined procedure of cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (HIPEC). Eur J Surg Oncol 2010; 36: 463–469. 35 González-Moreno S, Sugarbaker PH. Right hemicolectomy does not confer a survival advantage in patients with

mucinous carcinoma of the appendix and peritoneal seeding. Br J Surg 2004; 91: 304–311. 36 Youssef H, Newman C, Chandrakumaran K, Mohamed F, Cecil TD, Moran BJ. Operative findings, early complications, and long-term survival in 456 patients with pseudomyxoma peritonei syndrome of appendiceal origin. Dis Colon Rectum 2011; 54: 293–299. 37 Villeneuve L, Isaac S, Glehen O, Capovilla M, Chevallier A, Croce S et al. The RENAPE network: towards a new healthcare organization for the treatment of rare tumors of the peritoneum. Description of the network and role of the pathologists. Ann Pathol 2014; 34: 4–8.

Supporting information

Additional supporting information may be found in the online version of this article: Appendix S1 Statistical analysis (Word document) Table S1 Characteristics of 33 centres using cytoreductive surgery and intraperitoneal chemotherapy to treat pseudomyxoma peritonei (Word document) Table S2 Predictors of early oncological failure among 2451 patients with pseudomyxoma peritonei treated with cytoreductive surgery and intraperitoneal chemotherapy (Word document)

© 2014 BJS Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

BJS

Multicentre study of the learning curve and surgical performance of cytoreductive surgery with intraperitoneal chemotherapy for pseudomyxoma peritonei.

The learning curves for cytoreductive surgery with intraperitoneal chemotherapy for treatment of pseudomyxoma peritonei (PMP) were explored between in...
559KB Sizes 1 Downloads 8 Views