Content uploaded by Julia Pye
Author content
All content in this area was uploaded by Julia Pye on Apr 12, 2020
Content may be subject to copyright.
Oesophageal and gastric cancer pathology
reporting: a regional audit
S H Burroughs, AHBBiYn, J K Pye, G T Williams
Abstract
Aim—To audit the information content of
pathology reports of oesophageal and gas-
tric cancer resection specimens in Wales.
Methods—All such reports from the 16
NHS histopathology laboratories in Wales
in a one year period were evaluated for
their information content. Two standards
were used: (1) best practice reporting, and
(2) a minimum dataset required for
informed patient management that in-
cluded clear statements on histological
tumour type, depth of tumour invasion,
lymph node involvement, and complete-
ness of excision.
Results—282 reports were audited. Mini-
mum standards were achieved in 77% of
gastric resections (156/203) and 53% of
oesophageal resections (42/79). All labora-
tories achieved minimum standards in
some gastric cancer reports (range 50–
100%); three laboratories did not achieve
minimum standards in any oesophageal
cancer reports (range 0–100%). Best prac-
tice reporting was achieved in only 20% of
gastric and 18% of oesophageal cancer
reports. Failure to include an explicit
statement on completeness of excision or
involvement of the oesophageal circum-
ferential resection margin were the most
frequent causes of inadequate reporting.
Most other data items were generally well
reported, but apparent inadvertent omis-
sion of just one item was noted in many of
the substandard reports.
Conclusions—This audit shows the need
to improve the information content of
pathology reports in gastric and oesopha-
geal cancer. The widespread implementa-
tion of template proforma reporting is
proposed as the most eVective way of
achieving this. Multidisciplinary meetings
of clinicians involved in cancer manage-
ment should provide a forum for greater
communication between pathologists and
surgeons, and help to maintain standards
of pathological practice.
(J Clin Pathol 1999;52:435–439)
Keywords: audit; pathology reporting; gastric cancer;
oesophageal cancer
The overall five year survival for patients with
gastro-oesophageal cancer remains poor, at 5%
or less.12Only patients undergoing potentially
curative resection have any prospect of long
term survival, and in these, subsequent prog-
nosis is strongly linked to tumour stage.3–5 His-
topathological assessment of the resection
specimen plays a vital role in patient manage-
ment, in confirming whether complete excision
has been achieved and in providing essential
information for pathological TNM staging. As
well as contributing to the management of
individual patients, pathology reports provide
information for cancer registration, for clinical
audit, and for assessing the accuracy for new
diagnostic and preoperative staging tech-
niques, such as endoluminal and laparoscopic
ultrasound, contrast enhanced and spiral com-
puterised tomography, and magnetic reso-
nance imaging.
Guidelines for standardised surgical pathol-
ogy reports are published in histopathology
textbooks,6but such proforma based reporting
is not in widespread practice in the United
Kingdom. A recent audit of colorectal cancer
pathology reporting in Wales has highlighted
deficiencies in a significant number of reports,
with only 78% of colonic carcinoma reports
and 47% of rectal carcinoma reports meeting
minimum standards required for informed
patient management.7The results of this and
other audits have prompted the Royal College
of Pathologists to establish minimum datasets
for reporting a range of common cancers in
order to ensure that clinically important infor-
mation is recorded by histopathologists, and
these are in preparation for oesophageal and
gastric cancer. There is no published infor-
mation on the current quality of pathological
reporting in gastro-oesophageal carcinoma in
the United Kingdom. This paper presents the
results of an audit of histology reports of gastric
and oesophageal cancer resections in all of the
NHS laboratories in one United Kingdom
region (Wales) over a one year period.
Table 1 Data items abstracted from pathology reports and
used for audit
Criteria for best practice reports
Length of the specimen
Length of specimen along greater and lesser curves (gastric
resections)
Size of the tumour
Distance of tumour from the nearest resection end
Gross appearance of tumour
Histological tumour type
Degree of tumour diVerentiation
Lauren classification (gastric resections)
Extent of tumour spread (mucosa, muscle wall, beyond
muscle, reaching serosa)
Involvement of the resection ends by tumour (oesophageal
resections)
Involvement of the circumferential resection margin by tumour
(oesophageal resections)
Completeness of local excision (gastric resections)
Background abnormalities
Involvement of lymph nodes by tumour
Items abstracted but not included in best practice standards
Number of lymph nodes examined
Number of lymph nodes positive
Division of lymph nodes by distance (< or > than 3 cm) from
tumour (gastric resections)
J Clin Pathol 1999;52:435–439 435
Department of
Pathology, University
of Wales College of
Medicine, Heath Park,
CardiVCF4 4XN,
Wales, UK
S H Burroughs
G T Williams
Clinical EVectiveness
Support Unit (Wales),
Llandough Hospital,
Penarth, Vale of
Glamorgan, Wales, UK
AHBBiYn
Department of
Surgery, Wrexham
Maelor Hospital,
Wrexham, Clwyd,
Wales, UK
JKPye
Correspondence to:
Professor Williams.
email: WilliamsGT@cf.ac.uk
Accepted for publication
21 January 1999
on April 12, 2020 by guest. Protected by copyright.http://jcp.bmj.com/J Clin Pathol: first published as 10.1136/jcp.52.6.435 on 1 June 1999. Downloaded from
Methods
All cases of gastric and oesophageal cancer
treated in NHS hospitals in Wales between 1
September 1995 and 31 August 1996 were
included in a comprehensive audit, the remit of
which included clinical management and
follow up at one year, as well as histopathology
reporting of resection specimens. The clinical
director of each participating histopathology
laboratory was contacted to request permission
for inclusion of their departmental reports in
the study.Copies of all gastric and oesophageal
cancer resection pathology reports were subse-
quently obtained. The presence or absence of a
statement on items of information in these
reports was recorded on a proforma by three
experienced data collectors, and transferred to
a computer database by an optical mark scan-
ner to be analysed using the Statistical Package
for Social Sciences (SPSS for Windows). The
accuracy of data recording was validated by a
histopathologist checking the abstracted infor-
mation from a randomly selected 10% of
resections against the original pathology
report.
Table 1 shows the pathology information
items included in the proforma. These items
were selected following discussions at local
multidisciplinary upper gastrointestinal cancer
meetings, and on the basis of guidelines
presented in standard histopathology reference
books.6Reports which contained information
on all items were considered to satisfy best
practice standards. A subset of data,considered
to represent the minimum information neces-
sary for an adequate report, is shown in table 2.
To date, no nationally agreed standards for
reporting gastric and oesophageal cancer
specimens exist in the United Kingdom, and
the audit criteria were not circulated to partici-
pating pathologists before the period of data
collection. The results therefore reflect report-
ing practice before any interventions aimed at
quality improvement.
Results
In all, 282 pathology reports were included in
the audit, 72% of which were resections for
gastric cancer and 18% for oesophageal cancer.
Figure 1 shows the number of reports obtained
from each laboratory, which varied from three
to 34. Gastrectomy specimens were received by
all 16 laboratories; oesophageal resections were
submitted to 14 laboratories. Resections per-
formed for pathology other than carcinoma
(four non-Hodgkin lymphomas, one gastric
stromal sarcoma, one oesophageal melanoma)
were excluded from the study. Two further
resections were excluded from the analysis as
no residual local tumour could be identified in
the resection specimen following preoperative
chemoradiotherapy.
Validation of the abstracted data by a
pathologist in a random sample (over 10%) of
the reports showed a high level of agreement
with the interpretation of the original report for
the majority of information items. Occasional
interpretative errors were identified with re-
spect to gross appearance (diVusely infiltrating
tumours) and diVerentiation (anaplastic tu-
mours), leading to a slight underestimation of
the frequency of reporting of these two items in
the final results. However, interpretation of
items necessary for minimum acceptable re-
porting standards was accurate. Data items
regarding the total number of lymph nodes
examined and the number of positive nodes
were excluded from best practice criteria, as
such data were not standardised in the reports
and were subject to interobserver error when
abstracted by the data collectors. For example,
some laboratories received nodes, or groups of
nodes, as multiple separate surgical specimens
which were then reported individually. The
data item regarding identification of positive
nodes in gastrectomies as being within or
beyond 3 cm of the tumour was also excluded
from the best practice standard, as it is no
longer a requirement of the TNM staging
system.8
Table 2 Minimum criteria for an adequate report
Histological tumour type
Extent of tumour spread (mucosa, muscle wall, beyond
muscle, reaching serosa)
Involvement of the resection ends by tumour (oesophageal
resections)
Involvement of the circumferential resection margin by tumour
(oesophageal resections)
Completeness of local excision (gastric resections)
Involvement of lymph nodes by tumour
Figure 1 Number of reports obtained from each laboratory.
35
30
25
20
15
10
5
0
Laboratory
Number of reports
1
3
2
6
3
7
4
9
5
12
6
13
7
13
8
13
9
16
10
18
11
19
12
21
13
32
14
33
15
33
16
34
Table 3 Percentage of all gastric resection reports containing
statements on individual data items audited (all Wales)
Length of specimen 99.5
Distance along greater/lesser curves 83.3
Tumour size 88.2
Distance from nearest resection end 86.7
Gross appearance of tumour 86.7
Histological tumour type 97.0
Degree of tumour diVerentiation 83.3
Lauren classification 52.7
Extent of invasion 97.5
Completeness of local excision 83.7
Background abnormalities 73.4
Involvement of lymph nodes 98.0
436 Burroughs, BiYn, Pye, et al
on April 12, 2020 by guest. Protected by copyright.http://jcp.bmj.com/J Clin Pathol: first published as 10.1136/jcp.52.6.435 on 1 June 1999. Downloaded from
The overall percentage of reports containing
statements on each of the individual data items
included in the study is shown in tables 3 and
4. No item was present in every report,
although specimen length, tumour type, depth
of invasion, and the presence or absence of
lymph node involvement were recorded in the
vast majority of cases (> 95% of all reports).
Involvement of the proximal and distal ends of
the oesophageal resections was well reported
(96.2%), but the circumferential resection
margin was often neglected, being mentioned
in less than two thirds of cases. A clear
statement regarding local excision of gastric
tumours was missing in nearly one sixth of
reports. Background abnormalities were re-
corded more often in the stomach than in the
oesophagus (73.4% and 48.1%, respectively).
The poor recording of Lauren classification in
gastric resections was partly due to the
frequent use of the terms “signet cell” or “sig-
net ring” carcinoma when referring to the dif-
fuse form of gastric cancer described by
Lauren.9
The percentage of reports from individual
laboratories reaching minimum acceptable and
best practice reports are shown in table 5, along
with the results for Wales as a whole. These
results must be interpreted with caution in view
of the small numbers of resections performed
in some centres. Minimum standards were
reached in only 77% of gastric resections and
53% of oesophageal specimens. All laborato-
ries reached minimum standards in some
gastric reports but only one laboratory did so
consistently. Three of the 14 laboratories
reporting oesophageal resections failed to
reach minimum standards in any reports;
again, only one laboratory included the mini-
mum data set in 100% of reports. Best practice
reporting in a percentage of cases was achieved
by eight of 14 laboratories reporting oesopha-
geal cancers and 12 of 16 laboratories for gas-
tric resections. There was no correlation
between the number of gastro-oesophageal
resections received by individual laboratories
and the percentage of reports reaching mini-
mum standards (Spearman rank correlation
coeYcient, ñ= -0.29, p > 0.10). In addition,
good performance in gastric reporting by an
individual laboratory did not correlate with
good performance in oesophageal reporting
(fig 2).
Table 4 Percentage of all oesophageal resection reports
containing statements on individual data items audited (all
Wales)
Length of specimen 98.7
Tumour size 96.2
Distance from nearest resection end 84.8
Gross appearance of tumour 81.0
Histological tumour type 98.7
Degree of tumour diVerentiation 79.9
Extent of invasion 96.2
Resection end involvement 96.2
Circumferential resection margin involvement 62.0
Background abnormalities 48.1
Involvement of lymph nodes 92.4
Table 5 Percentage of reports fulfilling standards
Best practice standards
(table 1)
Minimum standards
(table 2)
Laboratory Oesophagus Stomach Oesophagus Stomach
A00 050
B2544 7556
C3311 8956
D1324 3864
E3832 6364
F 0 31 67 69
G 9 10 38 76
H0 0 080
I 11 0 33 83
J023 085
K–0 –86
L 0 25 75 88
M–11 –89
N 33 22 100 90
O3018 6091
P 0 18 50 100
All Wales 18 20 53 77
Figure 2 Percentage of reports from individual laboratories reaching minimum standards.
67 69
100
90
80
70
60
50
40
30
20
10
0
Laboratory
Oesophagus
Percent reports
A
50
000
B
75
56
C
89
56
D
38
64
E
63 64
P
50
100
O
60
91
N
100
90
L
75
88
J
85
I
33
83
H
80
G
38
76
F
Stomach
Oesophageal and gastric cancer pathology reporting 437
on April 12, 2020 by guest. Protected by copyright.http://jcp.bmj.com/J Clin Pathol: first published as 10.1136/jcp.52.6.435 on 1 June 1999. Downloaded from
The total number of lymph nodes examined
and the number of positive lymph nodes were
recorded in 67% of cases. The median number
of nodes examined in oesophageal resections
was 6 (range 1 to 19), and in gastric resections,
8 (range 1 to 36).
Discussion
High quality pathology reporting of cancer
resection specimens is essential for manage-
ment of individual patients, for establishing the
eYcacy of new preoperative staging techniques
and adjuvant treatment, for cancer registration,
and for organisation of cancer services. It is
therefore of concern that in our audit, only 77%
of gastric cancer reports and 53% of oesopha-
geal cancer reports satisfied the minimum
standards for an adequate report. The data
items selected for minimum standards—
tumour type, local excision, depth of invasion,
and involvement of lymph nodes—comprise the
basic information necessary for clinical man-
agement and tumour staging. Completeness of
local excision of tumour is the critical factor in
defining potentially curative surgery, and the
single most important prognostic parameter for
individual patients undergoing surgical
resection.5A clear statement regarding com-
pleteness of local excision was absent from a
significant proportion (12.7%) of gastrectomy
specimen reports, and was the most commonly
omitted minimum standard data item. The rea-
son for this is uncertain, though it may in part
be the result of the design of the proforma used
to abstract the information from original
pathological reports. Thus although a report
may have included (in the pathologist’s opin-
ion) all the information required to deduce
whether resection was complete (that is, depth
of tumour invasion and status of longitudinal
resection margins), such reports would not sat-
isfy the minimum standard criteria unless a
specific, unambiguous statement regarding
completeness of excision could be identified by
the data collectors. A wide range of clinical
staV, including surgeons, physicians, oncolo-
gists, palliative care teams, and general practi-
tioners may need to interpret pathology reports
on cancer resection specimens, and such
reports should therefore be readily comprehen-
sible to the non-histologist. For oesophageal
specimens, involvement of the proximal and
distal resection margins was almost always
commented upon (96.2%) but the status of the
circumferential resection margin was frequently
ignored, being included in only 62% of reports.
A retrospective study of oesophageal cancer
resections has shown that, as in the rectum, cir-
cumferential resection margin involvement by
tumour is related to subsequent local recur-
rence and decreased survival.10 Pathologists
clearly need to be aware of advances in clinical
practice, and be prepared to adapt their report-
ing of resection specimens accordingly. Regular
communication between pathologists and sur-
geons at multidisciplinary cancer management
meetings should provide an opportunity to
identify diYculties at a local level, and ulti-
mately improve and maintain quality of patho-
logical practice.
Achievement of minimum reporting stand-
ards as defined by this audit does not require
specialised training or interest in gastro-
intestinal disease, and should be within the
capacity of all diagnostic histopathologists.
Indeed, analysis of the individual data items
required to meet minimum standards shows
that tumour type, depth of invasion, and nodal
involvement were generally well reported
(97.5%, 97.2%, and 96.5%, respectively, for
gastric and oesophageal resections combined),
and that omission of these items from reports
is infrequent and inadvertent. Likewise, al-
though less than 20% of reports overall met
best practice reporting standards, only five of
the 23 individual items audited were recorded
in less than 80% of reports. These results sug-
gest that inconsistency in reporting, rather
than persistent poor performance, is responsi-
ble for the majority of substandard reports. In
this respect, our results are very similar to the
findings of the audit of colorectal cancer
reporting, also in Wales, published by Bull et
al.7There is no reason to believe that perform-
ance of Welsh NHS laboratories is any worse
or better than elsewhere in the United
Kingdom, although we are not aware of any
comparable published audit data for gastric
and oesophageal cancer reporting in other
NHS regions.
Several investigators have proposed the
implementation of standardised pathology pro-
formas as a means of ensuring consistent
incorporation of essential pathological data
into all reports.711 The eVectiveness of tem-
plate proformas in improving the information
content of mastectomy specimen reports and
colorectal cancer resection reports has recently
been demonstrated.12 13 Clinicians may also
find standardised reports easier to interpret
and extract information from. In addition,
template proformas facilitate computerised
data recording and retrieval for use in audit,
clinical trials, and cancer registration. It is
noteworthy that all the pathology reports
included in this audit were composed in free
form text. Reluctance of histopathologists to
use preprinted proforma reports may stem
from a perception that proformas are more
time consuming to complete than free text
reports, or are less aesthetically pleasing to the
pathologist compared with composing sen-
tences and paragraphs. However, we believe
that the argument in favour of widespread use
of template proformas for histopathology
reporting of cancer resections is convincing.
The construction of minimum datasets for
oesophageal and gastric cancer reporting, with
emphasis on evidence based practice, should
provide a framework for devising template pro-
formas which are acceptable to both patholo-
gists with ever increasing workloads and
clinicians in need of accurate and complete
prognostic information.
We wish to thank all of the pathologists and staVof the 16 NHS
histopathology laboratories in Wales who have kindly provided
access to the pathology reports, and Rita Carter and Kathy
Morris who collected the data from individual hospitals. We also
thank the Clinical EVectiveness Support Unit (Wales) for data
processing facilities and the Welsh OYce for financial support.
438 Burroughs, BiYn, Pye, et al
on April 12, 2020 by guest. Protected by copyright.http://jcp.bmj.com/J Clin Pathol: first published as 10.1136/jcp.52.6.435 on 1 June 1999. Downloaded from
Some of the data presented in this paper have been published in
abstract form (J Pathol 1998;186(suppl):9A).
1 Allum WH, Powell DJ, McConkey CC, et al. Gastric cancer:
a 25 year review.Br J Surg 1987;74:715–20.
2 Oliver SE, Rober tson CS, Logan RFA. Oesophageal cancer:
a population-based study of survival after treatment. Br J
Surg 1992;79:1321–5.
3 Sue-Ling HM, Johnston D, Martin IG, et al. Gastric cancer:
a curable disease in Britain. BMJ 1993;307:591–6.
4 Skinner DB, Ferguson MK, Soriano A, et al. Selection of
operation for esophageal cancer based on staging. Ann Surg
1986;204:391–401.
5 Hermanek P, Gospodarowicz MK, Henson DE, et al. Prog-
nostic factors in cancer. Berlin: Springer, 1995:37–63.
6 Rosai J. Ackerman’s surgical pathology, 8th ed. St Louis: CV
Mosby, 1996.
7 Bull AD, BiYn AHB, Mella J, et al. Colorectal cancer
pathology reporting: a regional audit. J Clin Pathol
1997;50:138–42.
8 Sobin LH, Wittekind C, eds. TNM classification of malignant
tumours, 5th ed. New York: John Wiley and Sons, 1997:60.
9 Lauren P. The two histological main types of gastric cancer.
Acta Pathol Microbiol Scand 1965;64:31–49.
10 Sagar PM, Johnston D, McMahon MJ, et al. Significance of
circumferential resection margin involvement at
oesophagectomy for cancer. Br J Surg 1993;80:1386–8.
11 Rosai J, members of Department of Pathology, Memorial
Sloan-Kettering Cancer Center. Standardized reporting of
surgical pathology diagnoses for the major tumour types. A
proposal. Am J Clin Pathol 1993;100:240–55.
12 Appleton MAC, Douglas-Jones AG, Morgan JM. Evidence
of eVectiveness of clinical audit in improving histopathol-
ogy reporting standards of mastectomy specimens. J Clin
Pathol 1998;51:30–3.
13 Cross SS, Feeley KM, Angel CA. The eVect of four
interventions on the information content of histopathology
reports of resected colorectal carcinomas. J Clin Pathol
1998;51:481–2.
The Royal College of Pathologists
“The Best” of the College Symposia
Academic activities, Continuing Professional Development, and Blackwell Healcare Commu-
nications are publishing a CD-ROM containing highlights of past College Symposia.
This CD-ROM is approved for CPD credits and is of general interest to pathologists. It
covers a wide range of topical issues, with lectures entitled:
xAllergy and asthma
xPeanut allergy
xA practical approach to unexpected infant death
xVitamin K prophylaxis: can we ever reach a consensus?
xEpidemiology of adverse eVects of “designer drugs”
xMulti-drug resistance modifiers: an alternative approach
xIdiopathic myelofibrosis: pathogenesis to treatment
xAdhesion molecules in pathology
xChromosomal changes and cancer
xInterpretation of injury in road traYc accidents
xReporting of cervical biopsies in the context of the cervical screening programmes
xThe metastatic process: its biological basis
xHuman papilloma virus infection
xThe hepatitis virus
xAlcoholic liver disease
The CD-ROM will cost you £30 for a personal copy or £120 for an institutional copy
with multiple user licence. To order a copy, call 0171 930 5862.
Oesophageal and gastric cancer pathology reporting 439
on April 12, 2020 by guest. Protected by copyright.http://jcp.bmj.com/J Clin Pathol: first published as 10.1136/jcp.52.6.435 on 1 June 1999. Downloaded from
