Role of circumferential margin involvement in the local recurrence of rectal cancer

8

Saurez ER, Rodemaker D, Hasson A, Mangogna L. High-dose steroids in childhood acute idiopathic thrombocytopenia purpura.

13 Blanchette

Am J Pediatr Hematol/Oncol 1986; 8: 111-15. 9

Jayabose S, Patel P, Inamder S, et al. Use of intravenous methylprednisolone in acute idiopathic thrombocytopenic purpura.

10

van

Semin Hematol 1992; 29: 77-82. 14 Bussel JB, Pham LC. Intravenous treatment with gammaglobulins in adults with immune thrombocytopenic purpura: review of the literature. Vox Sang 1997; 52: 206-11. 15 Imbach P, Wagner HP, Berchtold W, et al. Intravenous immunoglobulin versus oral corticosteroids in acute immune thrombocytopenic purpura in childhood. Lancet 1985; ii: 464-68. 16 Salama A, Kiefel V, Amberg R, Mueller-Eckhardt C. Treatment of autoimmune thrombocytopenic purpura with rhesus antibodies (antiRho [D]). Blut 1984; 49: 29-35. 17 Andrew M, Blanchette VS, Adams M, et al. A multicentrer study of the treatment of childhood chronic idiopathic thrombocytopenic purpura with anti-D. J Pediatr 1991; 120: 522-27.

Am J Pediatr Hematol/Oncol 1987; 9:

133-35.

Hoff J, Riutchey AK. Pulse methylprednisolone therapy for acute childhood idiopathic thrombocytopenic purpura. J Pediatr 1988; 113: 563-66.

11

12

Özsoylu S, Irken G, Karabent A. High-dose intravenous methylprednisolone for acute childhood idiopathic thrombocytopenic purpura. Eur J Haematol 1989; 42: 431-35. Imbach P, Barandum S, d’Apuzzo V, et al. High-dose intravenous gammaglobulin for idiopathic thrombocytopenic purpura in childhood. Lancet 1981; i: 1228-30.

VS, Kirby MA, Turner C. Role of intravenous immunoglobulin G (IVIgG) in autoimmune hematologic disorders.

Role of circumferential margin involvement in the local recurrence of rectal cancer

Summary

Introduction

Local recurrence after resection for rectal cancer remains common despite growing acceptance that inadequate local excision may be implicated. In a prospective study of 190 patients with rectal cancer, we examined the circumferential margin of excision of resected specimens fortumour presence, to examine its frequency and its relation to subsequent local

Local recurrence of rectal cancer after surgery is common and probably influences survival. The frequency of local recurrence varies according to the treating surgeon. 1,2 Several mechanisms have been postulated as the cause of local recurrence, including the implantation of exfoliated tumour cells at the anastomosis during resection and the promotional effects of the anastomosis on tumour growth.4 However, these factors do not explain that in many cases tumour recurrence is not mucosal but intramural or extramuralthe difference in frequency of local recurrence in rectal and colonic cancer, or the variation of frequency between surgeons. In a series of 52 patients we found that tumour involvement of the circumferential resection margin was a powerful predictor of subsequent local recurrence.6 We concluded that most local recurrences occurred as a direct result of inadequate tumour resection and that this factor may explain variation in local recurrence rates between

recurrence.

Tumour involvement of the circumferential margin was seen in 25% (35/141) of specimens for which the surgeon thought the resection was potentially curative, and in 36% (69/190) of all cases. After a median 5 years’ follow-up (range 3·0-7· 7 years), the frequency of local recurrence after potentially curative resection was 25% (95% Cl 18-33%). The frequency of local recurrence was significantly higher for patients who had had tumour involvement of the circumferential margin than for those without such involvement (78 [95% Cl 62-94] vs 10 [4-16]%). By Cox’s regression analysis tumour involvement of the circumferential margin independently influenced both local recurrence (hazard ratio=12·2 [4·4-34·6]) and survival (3·2 [1·6-6·53]). These results show the importance of wide local excision during resection for rectal cancer, and the need for routine assessment of the circumferential margin to assess

prognosis.

surgeons.l,2 Since that study6 support for wide local surgical clearance of rectal cancer has grown. However, two other studies have yielded conflicting results: one study7 did not confirm the association with local recurrence, although it found a significant influence on survival, whereas the other confirmed the results in 80 patients by meticulous wholemount sections, but could only present data based on 2 years’ follow-up. In our study, 190 patients were followed up for a median of 5 years with the aim of clarifying the importance of complete local excision at the circumferential

margin. Patients and methods October, 1985, and June, 1990, specimens from all rectal patients who had had resection at the Leeds General Infirmary were examined by pathologists who routinely recorded the Dukes’9 and JasslO stages and various pathological variables. Examination for tumour at the circumferential resection margin has been reported in detail." It involved slicing the resected specimen transversely to provide multiple coronal sections through Between cancer

Academic Units of Surgery (I J Adam FRCS, I G Martin FRCS, PJ Finan MD Prof D Johnston MD) and Pathological Sciences (M O Mohamdee MRCPath, N Scott MRCPath, M F Dixon MD, P Quirke PhD), Centre for Digestive Diseases, The General Infirmary at Leeds, Leeds LS1 3EX, UK

Correspondence to: Dr P Quirke

the tumour and the associated mesorectum. The mesorectum above the tumour was also coronally sliced from the posterior aspect and examined for deposits of tumour at the circumferential margin. 3 or 4 tissue blocks were taken where tumour approached

707

Table 1:

Operative details

closest to the margin. We measured the closest point of the tumour to the circumferential margin microscopically, and any specimen that had tumour 1 mm or less from the circumferential margin of excision was recorded as having tumour involvement. In July, 1993, one researcher (IJA) independently reviewed the patients’ clinical details and subsequent outcome. Any patient who had had preoperative radiotherapy was excluded from the trial because of reports that it may downstage the tumour." Information about patient survival, cause of death, and local recurrence was retrieved from patients’ notes and general practitioner records. All patients were followed up in general surgical clinics. Only one consultant specifically and routinely looked for local recurrence in the absence of symptoms. All other consultants investigated only patients with symptoms. The surgeon who routinely looked for local recurrence found a lower frequency of local recurrence than that seen overall in the study. The presence or absence of information about circumferential resection margin status did not affect management. All information was cross-checked against the death certificate and files of the Yorkshire Regional Cancer Organisation. Local recurrence was defined as any recurrence within the pelvis and was recorded only when confirmed by positive histology or by radiographic imaging. Cancer-specific survival was determined in all cases. Information was collected from the hospital notes and the individual notes of the surgeon to identify whether the original resection was palliative or potentially curative. A potentially curative resection was defined as one in which all macroscopic tumour tissue had been excised and there was no evidence of distal margin involvement.

v

’-’

i

Follow up

...

(years)

1: Cumulative frequency of local recurrence comparing presence or absence of tumour at circumferential resection margin (CRM) In patients who had potentially curative

Figure

resection

cancer-specific survival. Data for the influence of tumour involvement of the circumferential resection margin on overall survival on all patients, and its effects on survival and the cumulative risk of local recurrence was tested with the log-rank method. 13 We then analysed the data to find out how each pathological variable influenced either the cumulative risk of local recurrence or survival by log-rank method. Variables that had a significant effect were compared with Cox’s regression analysis14 to determine which had an independent effect on prognosis.

Statistical analysis

Results

Data

patients (104 male, 86 female) with a median age of 69 years (range 32-88) were enrolled in our prospective study between October, 1985, and June, 1990. 6 patients were excluded from analysis-3 received preoperative 3 lost to and were follow-up. The median radiotherapy 5-3 was years (3-0-7-7 years). The patients were follow-up operated on by 23 different surgeons, and the operative procedures used are shown in table 1. In 141 patients (74%) the operation was potentially curative. Residual tumour remained of the pelvis at time of operation in 23 patients (12%). Operative mortality (30 day) was 6% (11 patients) and was greater in patients who had a palliative procedure. 11 patients received postoperative radiotherapy (6%).

the

variables plus survival and local entered into the computer programme Database III (Ashton-Tate). For the purpose of the analysis, the following pathological variables were used: tumour penetration through the muscularis propria, penetration of the serosa, presence of involved lymph nodes, presence of distant metastases, involvement of the apical lymph node, histological grade, presence of a lymphocytic infiltrate, appearance of the tumour border, presence of extramural venous invasion, involvement of the distal margin of excision by tumour, involvement of the circumferential margin of excision by 1 mm). tumour (involved 1 mm, or clear > The initial analysis examined the influence of circumferential margin involvement by tumour on local recurrence in all patients in the study, irrespective of whether they had had a palliative or potentially curative resection. This analysis was repeated for both on

pathological

recurrence were

groups In a

separately. subsequent analysis of survival, patients who

died from non-cancer-related illness were censored from further analysis from the time of death. All results are expressed in terms of

CRM =circumferential resection margin

Table 2: Local recurrence: influence of circumferential resection (CRM) involvement by tumour

708

margin

190

Tumour involvement of circumferential resection margin Involvement of the circumferential resection margin was found in 69 specimens (36%). 35 (25%) of the 141 patients who were considered to have had potentially curative resections had involvement of the circumferential margin. Local recurrence Local recurrence was diagnosed in 55 patients (29%). Of the 69 patients who had tumour involvement of the circumferential margin, 44 had subsequent local recurrence, whereas recurrence developed in only 11 patients without such involvement. Local recurrence developed in 32 of 141 patients who had a potentially curative resection (23%) and in 23 of 49 who had a palliative procedure (47%). For patients with a clear circumferential margin who had potentially curative resection, the percentage without local recurrence at 5 years was 90%

-

v

Follow up

Follow up (years)

Figure 2: Cumulative frequency of proportion of patients with no local recurrence comparing presence or absence of tumour at circumferential resection margin (CRM) In patients who underwent potentially curative resection for Dukes’ B or C tumours

(95% CI 84-96); patients with tumour at the circumferential margin did significantly worse, with a cumulative percentage without recurrence at 5 years of only 22% (6-38, log rank p < 0001, figure 1). Figure 2 shows the life tables for cumulative percentage without local recurrence for patients who had a potentially curative resection for Dukes’ B or Dukes’ C tumour. Of the 55 patients who had local recurrence, 7 had a second

operative procedure (13%). Survival The cancer-specific overall 5-year survival was 48% (40-56). Figure 3 shows the life table for survival by Dukes’ stage (5-year survival Dukes’ A 86%, B 64%, C 40%, Cl 44%, C2, 23%). The 5-year survival percentage in the 69 patients with tumour involvement at the circumferential margin was 15% (6-25), compared with 66% (57-75) in those without such involvement (log rank p < 0001). For

-

_

(years)

Figure 4: Survival comparing presence or absence of tumour at the circumferential resection margin (CRM) In patients who underwent potentially curative resection the 141 patients who had a potentially curative resection, overall 5-year survival was 62% (53-70). Again, patients who had tumour involvement of the circumferential margin of excision (n = 35) had significantly poorer 5-year survival than those without such involvement (24 [8-39] vs 74 [65-83] %, log rank p < 0001). The life tables for these two groups are shown in figure 4.

Multivariate analysis To assess the effects of circumferential margin involvement in patients with no residual disease, the effect on outcome of each pathological variable was assessed for patients who had a potentially curative resection (table 3). Variables that had a significant result in this univariate analysis were then used in a multivariate analysis to see if they had an independent effect. We found that the same three variables were of independent significance to determine poor survival and a higher cumulative percentage of local recurrence: the

Studies done on

potentially curative resections only. : Influence of pathological variables

Table 3: Univariate analysis on

follow up (years) Survival 3: Dukes’ by stage Figure Proportion of cases: 13% A, 31% 8, 39% C, 17% D.

prognosis

Table 4: Multivariate analysis : pathological variables

Independently Influencing prognosis 709

presence of tumour at the circumferential margin of excision; lymph node involvement by tumour; and an infiltrating appearance of the tumour border (table 4). Patients with circumferential margin involvement were 3 times more likely to die and 12 times more likely to have local recurrence than patients without circumferential margin involvement, despite apparent complete clearance of tumour at operation.

Discussion The results of our prospective study support the hypothesis that the presence of tumour at the circumferential resection margin is important in determining the subsequent development of local recurrence in many patients. The risk of a local recurrence after a resection that was initially regarded as potentially curative was 12 times greater if the circumferential margin of excision was involved by tumour. Furthermore, along with lymph-node involvement and an infiltrating margin at the tumour edge, the presence of tumour at the circumferential resection margin is an important independent determinant of overall survival. The multivariate analysis suggests that for potentially curative resections, the critical factor is the involvement of the circumferential margin rather than penetration of the muscularis propria by tumour. The message for the surgeon is clear-wide local excision of a rectal cancer is important in determining a good prognosis for the patient. Although circumferential margin involvement was a powerful predictor of local recurrence, our study also shows that local recurrence develops in a few patients without circumferential resection margin involvement. Previous studies have looked at other mechanisms for the development of local recurrence in rectal cancer, such as free exfoliated tumour cells3 and the promotional effects of the anastomosis.4 These mechanisms may have brought about local recurrence in the patients without circumferential margin involvement. A further possibility is the presence of discontinuous local spread of tumour into an area of the mesorectum not removed by the surgeon.15 In our study some patients had circumferential resection margin involvement, but did not develop local recurrence. In most of these patients, widespread tumour dissemination was diagnosed during follow-up, but intensive investigation specifically to identify local recurrence was not indicated. Other possible reasons for the apparent non-progression to local recurrence are inadequate follow-up or the effects of postoperative adjuvant radiotherapy. A further possibility is the creation of an artifactual circumferential margin during removal or subsequent handling of the tumour. Our finding that the circumferential margin was involved by tumour in 25% of resections at which the surgeon had considered the margins of excision to be clear is important. The reported frequency of local recurrence varies widely from 6% in specialist units to more than 50%. Recurrence rates between 15 and 35% are not unusual in routine surgical practice, which has led to calls for rectal cancer to be treated by specialist surgeons only .16 We suggest that the variation of local recurrence may be due to the frequency of unsuspected involvement of the circumferential resection margin. In this study, 23 different surgeons were involved, no doubt with consequent variations in surgical technique, and there was a high frequency of local recurrence (23%) among patients who had potentially curative resection. Our results confirm those of Ng et al8 but not those of Cawthorn and colleagues.’ In that study,7 the frequency of local 710

8%: circumferential margin involvement was found in only 6-6% of specimens and did not predict local recurrence. However, the study was done by a small number of surgeons who all used the technique of total mesorectal excision; it has been suggested that such recurrence was

complete excision of the mesorectum may allow death from metastatic disease before local recurrence is detected." Concern over high rates of local recurrence has led to increasing interest in adjuvant radiotherapy for rectal cancer. Both preoperative and postoperative radiotherapy can reduce local recurrence rates.18,19 Our results suggest that any trial on the influence of adjuvant therapy on local recurrence that does not allow for unexpected tumour involvement of the circumferential margin of excision needs to be interpreted with caution. Abulafi and Williams20 have highlighted the need for improvements in both imaging techniques and routine histological examination when planning adjuvant therapy. The ability of the finding of involvement of the circumferential resection margin to predict the risk of local recurrence suggests it is a variable on which to base decisions about postoperative adjuvant radiotherapy. Indeed, we would question the need for any adjuvant therapy in the apparently excised Dukes’ B tumour when the circumferential margin is clear. Good surgical technique can result in cure in most of these patients. 15,22 Improvements in surgical technique that lead to a reduction in local recurrence may benefit the patient more than adjuvant therapy alone.1 The pathological method itself is designed to be simple and rapid. It adds no more than 10 min to the assessment of a rectal cancer, it is easy to teach, and is very cost-effective, adding no more than four blocks to a case. This work was supported by the Yorkshire Cancer Research Campaign and the Special Trustees of Leeds General Infirmary. Secretarial assistance was provided by Mrs J Feamly.

References 1

2

3 4

5 6

7

8

9

10 11

12 13

McArdle CS, Hole D. Impact of variability amongst surgeons in post-operative morbidity and mortality and ultimate survival. BMJ 1991; 302: 1501-05. Phillips RKS, Hittinger R, Blesovsky L, Fry JS, Fielding LP. Local recurrence following "curative" surgery for large bowel cancer: I, the overall picture. Br J Surg 1984; 71: 12-16. Gordon-Watson C. Origin and spread of cancer of the rectum in relation to surgical treatment. Lancet 1938; i: 239-45. Williamson RCN, Davis PW, Bristol JB, Wells M. Intestinal adaption and experimental carcinogenesis after partial colectomy: increased tumour yields are confined to the anastomosis. Gut 1982; 23: 316-25. Morson BC, Vaughn EG, Bussey HJR. Pelvic recurrences after excision of the rectum for carcinoma. BMJ 1963; 2: 13-18. Quirke P, Durdey P, Dixon MF, Williams NS. Local recurrence of rectal adenocarcinoma due to inadequate surgical resection: histopathological study of lateral tumour spread and surgical excision. Lancet 1986; ii: 996-99. Cawthorn SJ, Parums DV, Gibbs NM, et al. Extent of mesorectal spread and involvement of lateral resection margin as prognostic factors after surgery for rectal cancer. Lancet 1990; 335: 1055-59. Ng IOL, Luk ISC, Yuen ST, et al. Surgical clearance in resected rectal carcinomas; a multivariate analysis of clinicopathologic features. Cancer 1993; 71: 1972-76. Dukes CE. Classification of carcinoma of the rectum. J Pathol Bacteriol 1932; 35: 330-32. Jass LR, Love SB, Northover JMA. A new prognostic classification of rectal cancer. Lancet 1987; i: 1303-06. Quirke P, Dixon MF. How do I do it: the prediction of local recurrence in rectal adenocarcinoma by histopathological examination. Int J Colon Dis 1988; 3: 127-31. Powers WE, Tolmach LJ. Preoperative radiation therapy: biological basis and experimental investigations. Nature 1964; 201: 172-204. Peto R, Pike M. Conservation of the approximation (O-E)2/E in the log rank test for survival data or tumour recurrence data. Biometrics

1973; 29: 579-84.

14 Cox DR. 187-219.

Regression models and life tables. J Roy Stat Soc 1972; 34:

15 Heald RJ, Ryall RD. Recurrence and survival after total mesorectal excision for rectal cancer. Lancet 1986; i: 1479-82. 16

Fielding LP, Stewart-Brown S, Dudley HAF. Surgeon-related variables and the clinical trial. Lancet 1978; ii: 778-79.

Breaching the mesorectum. Lancet 1990; 335: 1067-68. Stockholm Rectal Cancer Study Group. Short-term preoperative radiotherapy for adenocarcinoma of the rectum. Am J Clin Oncol 1987;

17 Editorial. 18

10: 369-75.

19 Fisher B, Wolmark N, Rockette H,

et al. Post-operative adjuvant chemotherapy or radiation therapy for rectal cancer: results from NSABP Protocol R-01. J Natl Cancer Inst 1988; 80: 21-29.

20 Abulafi AM, Williams NS. Local recurrence of colorectal cancer: the problem, mechanisms, management and adjuvant therapy. Br J Surg 1994; 81: 7-19. 21 Chan KW, Boey J, Wong SKC. A method of reporting radial invasion and surgical clearance of rectal carcinoma. Histopathology 1985; 9: 1319-27. 22 MacFarlane JK, Ryall RDH, Heald RJ. Mesorectal excision for rectal cancer. Lancet 1993; 341: 457-60.

Geographical variations in plasma viscosity and relation to coronary event rates

Summary

Introduction

Plasma viscosity is reported to be predictive of coronary heart disease (CHD) and stroke. To find out whether regional differences in CHD event rates correlate with differences in plasma viscosity, we compared plasma viscosity in a high-risk area for CHD (Glasgow Multinational Monitoring of Trends and Determinants in Cardiovascular Disease [MONICA] and Scottish Heart Health Study population surveys, 1985/86; n=1166) and in a lower-risk area (MONICA Augsburg survey, 1984/85; n=3258) in men and women aged 25-64 years. Mean plasma viscosity (37°C) was 1·261 (SD 0·067) mPa s in Augsburg and 1·327 (0·093) mPa s in the west of Scotland for men, and 1·248 (0·066) mPa s and 1 318 (0·087) mPa s, respectively, for women. The unadjusted difference of the means between the west of Scotland and Augsburg was 0·066 (95% Cl from weighted regression 0·058-0 073) mPa s for men and 0·070 (0·062-0·078) mPa s for women. Adjustment for age, smoking behaviour, total and high-density-lipoprotein cholesterol, systolic and diastolic blood pressure, and body-mass index had no effect on these differences. Age-standardised coronary event rates in 1985-87 were at least two times higher among men, and four times higher among women, in MONICA Glasgow than in MONICA Augsburg. This large geographical difference in plasma viscosity might partly explain the differences in CHD event rates between these populations. Further studies are needed on the determinants of plasma viscosity, and on its potential roles in atherosclerosis, thrombosis, and ischaemia.

Epidemiological, pathophysiological, and experimental data strongly suggest a key role for the haemostatic system, especially fibrinogen, in the initiation of atherosclerosis and its clinical complications. One mechanism by which fibrinogen and other plasma proteins could promote coronary heart disease (CHD) is by increasing plasma viscosity, which is related to other important risk factors for CHD.2-8 In addition

conventional risk factors,2,5-7 plasma viscosity was investigated in southern Germany, in the first cross-sectional study of the Multinational Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA) Augsburg project; and in the west of Scotland, in the first MONICA survey in Glasgow and in the concurrent Scottish Heart Health Study. We describe here the distributions of plasma viscosity and its known correlates in these two population samples; we related these findings to CHD event rates, which at the time of these studies were high in the west of Scotland and lower in to

Augsburg. Methods The MONICA project is a World Health Organization (WHO) coordinated, observational long-term study. Its main objective is to measure trends in cardiovascular mortality and morbidity and to assess the extent to which these trends are related to changes in risk factor levels and medical care, measured at the same time in different communities in different countries.9,lo It consists of a coronary event register and three independent cross-sectional studies with 5-year intervals.

Augsburg MONICA survey Department of Internal Medicine II, Ulm University Medical Centre, Ulm (W Koenig MD); GSF Research Centre for Environment and Health MEDIS Institute, Neuherberg, Germany (M Sund MS); Department of Medicine, University of Glasgow, Royal Infirmary, Glasgow (Prof G D O Lowe FRCP); Cardiovascular Epidemiology Unit, Ninewells Hospital and Medical School, University of Dundee, UK (AJ Lee PhD, Prof H Tunstall-Pedoe FRCP); Department of Physical Medicine and Rehabilitation, University of Vienna, Vienna, Austria (K L Resch MD); Institute of Epidemiology and Social Medicine, University of Münster, Germany (Prof U Keil MD); and Postgraduate Medical School, University of Exeter, Exeter, UK (Prof E Ernst MD) Correspondence to: Dr W Koenig, Department of Internal Medicine II, University of Ulm Medical Centre, Robert-Koch-Str 8, D-89081 Ulm, Germany

The first survey in the Augsburg MONICA project (southern Germany) was carried out in 1984-85. 4022 of the 5312 randomly sampled eligible subjects, aged 25-64 years, took part (response rate 75%).11-13 This report is based on a subsample of 3258 men and women, in whom plasma viscosity was measured. There were no appreciable differences in conventional cardiovascular risk variables between the subgroup and the total sample."

Glasgow MONICA survey and Scottish Heart Health Study In the first Glasgow MONICA survey, undertaken during 1986, 984 men and women aged 25-64 years were randomly selected from general practitioners’ lists in North Glasgow District.9 The Scottish Heart Health Study (1985-86) was based on 10 359 men and women aged 40-59 years randomly selected from general

711