Consensus recommendations for the standardized histopathological evaluation and reporting after radical oesophago-gastrectomy (HERO consensus) Free

SUMMARY

Background

Variation in the approach, radicality, and quality of gastroesophageal surgery impacts patient outcomes. Pathological outcomes such as lymph node yield are routinely used as surrogate markers of surgical quality, but are subject to significant variations in histopathological evaluation and reporting. A multi-society consensus group was convened to develop evidence-based recommendations for the standardized assessment of gastroesophageal cancer specimens.

Methods

A consensus group comprised of surgeons, pathologists, and oncologists was convened on behalf of the Association of Upper Gastrointestinal Surgery of Great Britain & Ireland. Literature was reviewed for 17 key questions. Draft recommendations were voted upon via an anonymous Delphi process. Consensus was considered achieved where >70% of participants were in agreement.

Results

Consensus was achieved on 18 statements for all 17 questions. Twelve strong recommendations regarding preparation and assessment of lymph nodes, margins, and reporting methods were made. Importantly, there was 100% agreement that the all specimens should be reported using the Royal College of Pathologists Guidelines as the minimum acceptable dataset. In addition, two weak recommendations regarding method and duration of specimen fixation were made. Four topics lacked sufficient evidence and no recommendation was made.

Conclusions

These consensus recommendations provide explicit guidance for gastroesophageal cancer specimen preparation and assessment, to provide maximum benefit for patient care and standardize reporting to allow benchmarking and improvement of surgical quality.

Background

There is significant variation in the approach, extent, and quality of gastroesophageal surgery, which impacts on patient outcomes.¹ In response, efforts to standardize care and develop means of quality assurance and assessment of surgery have been applied.² The most commonly used measures of surgical quality relate to the pathological analysis derived from the excised specimen.

The number of lymph nodes harvested at surgery is commonly used as a measure of surgical quality, based upon studies which have demonstrated associations between extent of lymphadenectomy and patient survival.^3–5 Some surgeons, however, have raised significant concerns regarding the use of such metrics as a surrogate measure of quality. Chief among these criticisms is that variations in practices of pathology laboratories and histopathologists may result in significant differences in pathological outcomes,^6,7 meaning that comparisons between surgeons and surgical centers will not be possible.

A recent UK national audit of practice of assessment of esophagogastric cancer specimens conducted by the Association of Upper GastroIntestinal Surgery of Great Britain & Ireland (AUGIS), demonstrated the broad variation present across tertiary specialist centres.⁸ For example, three different definitions of margin positivity were commonly used across the country and only 16% of units met the minimum reporting recommendations set out in national guidance by the Royal College of Pathologists (RCPath). There was little or no standardization of practice in the assessment of the primary tumor, margins, or lymph node yield. Whereas some units counted only manually palpable lymph nodes, others systematically blocked all fatty tissue for microscopic assessment, resulting in significantly higher nodal counts.

Without standardization of practice in the assessment and reporting of surgical specimens, histopathological outcomes cannot be considered valid surrogates of surgical quality, or be used as benchmarks to compare care across units. A multidisciplinary, pan-stakeholder working group was therefore established on behalf of AUGIS to create guidelines for practice based on currently available evidence.

The aim of these consensus guidelines was to recommend best practices in the handling, preparation, assessment, and reporting of surgical specimens, following radical surgery for esophagogastric cancer. The resulting recommendations are intended to¹ optimize assessment of the primary tumor and associated tissues to guide patient treatment with reference to adjuvant therapy and prognosis, and² standardize reporting to allow surgical quality benchmarking across units.

METHODS

A multidisciplinary consensus group of surgeons, pathologists, and oncologists was convened including representation of AUGIS, within a consensus guidelines methodological framework.^9,10

Phase one

Using a modified nominal group technique methodology,¹¹ an electronic questionnaire was circulated to consensus group members to identify priority areas for consideration. The nominal group technique allows for anonymized contribution of ideas without risk of bias being introduced by the opinions of other group members. Resulting ideas are then collated and each idea’s inclusion or exclusion from the further process decided upon by majority vote, this resulted in 17 final topics (Table 1).

Phase two

A literature review was conducted for each topic. Priority was given to higher quality and higher levels of evidence, and strength of evidence rating assigned in keeping with the GRADE approach¹²; evidence summaries were circulated to all group members for review.

Phase three

A virtual consensus workshop was held on January 20, 2021. Each topic was debated and draft consensus statements developed based on the agreed evidence base and group expert opinion.

Phase four

Each statement was subjected to an anonymous consensus voting process (Delphi). Members were asked to vote on both their level of disagreement or agreement with each statement (Likert scale 1–5), and strength of recommendation (weak or strong). Consensus was considered achieved if >70% of respondents were in agreement. A single topic which did not achieve consensus was recirculated for a second round of voting. All topics achieved consensus by the second round.

RESULTS

How should the extracted specimen be managed with reference to opening of the specimen and/or fixation to a rigid backing?

It is impractical to expect pathology laboratories to receive and deal with these specimens fresh from theatre; they should be fixed in formalin solution after surgical excision. It is recommended that these specimens are not opened in theatre unless this is required e.g. to allow tissue sampling for research purposes—in which case a locally agreed protocol should be used. Specimens should be opened and cleaned in the pathology laboratory in such a way that minimizes disruption to the tumor and surgical margins.

Weak recommendation, very low certainty of evidence.

Specimens should be fixed to a rigid backing to prevent shrinkage and help with orientation in line with RCPath guidance.

Weak recommendation, very low certainty of evidence.

No published evidence specific to this topic was found; recommendation here is based upon expert consensus. Current RCPath guidance recommends delivery of fresh specimens to the pathology lab, or where this is not possible to consider opening of the specimen. The consensus group identified that in the majority of cases it is not practical for fresh specimens to be delivered; some centers have surgical and histopathological departments located at separate sites, and even where co-located, the majority of units do not have the resources to prepare specimens on arrival. The opening of specimens in theatre risks compromising the circumferential resection margin and specimen orientation. We therefore recommend that specimens be delivered non-opened and fixed, unless opened in an agreed and standardized fashion for research purposes which does not compromise pathological assessment. The risks of disturbing the specimen prior to pathologic assessment must be weighed against the potential research benefits, and should be clearly detailed as part of any study’s protocol.

Current RCPath guidance recommends that specimens be fixed (pinned or sutured) to a rigid backing prior to fixation, if not being delivered fresh to the pathology lab. This is in order to mitigate shrinkage associated with formalin fixation,¹³ which can distort the specimen, and starts as soon as the specimen is devascularised. However, evidence regarding the effect of shrinkage on assessment of the specimen is variable and of poor quality; a recent systematic review concluded that the available poor-quality evidence did not show that longitudinal margin length had any effect on prognosis or treatment.¹⁴

How long should the specimen be fixed prior to cut-up?

Specimens should be left to fix for at least 24 hours. Inking of non-peritonealised margins prior to opening through these will help the assessment of these during subsequent dissection and histopathological reporting.

Strong recommendation, very low certainty of evidence.

In the absence of specific evidence, recommendations are based on expert consensus. Limited study of the effect of prolonged (>96 hours) versus normal (24–48 hours) fixation in other tumor types has not found any difference in pathological or immunohistochemical analysis.¹⁵

Should additional dissection or node harvest from the specimen be performed by the surgical team prior to fixation of the specimen?

We recommend that the surgeon dissects distant nodal stations off the specimen before it is sent to the pathologist, but all peri-tumoural nodes should be left attached, with no dissection of the primary tumor, thus enabling adequate assessment of the CRM. Specific nodal stations which should be dissected (if separate to primary tumor) should include common hepatic artery nodes, left gastric artery nodes, splenic artery nodes, subcarinal nodes, right paratracheal nodes, and para-aortic (thoracic duct) nodes.

Strong recommendation, low certainty of evidence.

Nodal dissection by the surgeon can be beneficial in the identification of separate nodal stations or anatomical areas, as tissue shrinkage and the effects of formalin fixation can make this more challenging in fixed specimens. In the context of nodal yields being used as a surgical quality surrogate, there can also be greater motivation on the part of the surgeon to achieve greater harvest when this is performed on the fresh specimen by the surgical team. Limited evidence suggests that additional surgical dissection of lymph nodes before sending to the pathologist will increase number of nodes identified. A subset analysis of the multicenter (23 institutions) US Z0060 trial compared results for 134 specimens, wherein separate submission of nodal packets resulted in a higher nodal yield compared with complete specimens, though as a whole, nodal yield was low (median 16 vs. 10 lymph nodes).¹⁶ Similar results were reported by Cichowitz et al. in a retrospective single-center Australian study (median 19 vs. 6 lymph nodes),¹⁷ and an American study of gastrectomy specimens (median 40 vs. 27 lymph nodes).¹⁸ Cichowitz et al. also reported significant variations between reporting histopathologists.¹⁷

An elegant RCT in uterine cancer assessed this issue by randomizing patients to having one side of bilateral pelvic nodal dissection specimens sent as either a complete specimen, or divided into four nodal packets (common, external, internal iliac, and obturator), with the contralateral side serving as control.¹⁹ Specimens sent in divided tissue packets results in significantly higher nodal counts, independent of the assessing pathologist, though significant differences between pathologists’ median nodal counts were also noted.

Thus, we recommend dissection of individual nodal stations to standardize and improve nodal harvest, and to give additional information about the geography of nodal spread, as this will potentially give important prognostic information. As we move to more personalized oncological therapy the pattern of nodal involvement may guide the need for and type of adjuvant therapy.²⁰

There is a clear concern that if all the lymph nodes are dissected off the specimen, in removing peritumoural tissues the CRM is compromised and impossible to assess. CRM status remains one of the important surgical quality markers used by the National Oesophago-Gastric Cancer Audit (NOGCA), in addition to its importance for patient prognosis and consideration of adjuvant treatment.²¹ A UK study of 301 resection specimens with complete ex vivo surgical nodal dissection reported that as a result of the dissection, CRM was not able to be assessed in 20.6% of cases.²² Accordingly, we recommend that any tissues located close to the tumor or otherwise possibly involved in assessment of margins should be left intact for pathological assessment.

Who should tissue blocks (incl. tumor and lymph node identification) be cut-up/prepared by?

We recommend the responsible pathology consultant ensure the dissector is suitably competent and adheres to all the departmental standard operating practices and RCPath dataset guidelines. Each histopathology department should aim to have a named consultant who can take the lead on dealing with these specimens.

Strong recommendation, very low certainty of evidence.

A previous national survey from the HERO consensus group found that there was a mixed approach to specimen dissection with some units using consultants, trainees, or biomedical scientists (BMS) only and others using a combination of those personnels to examine and sample specimens.⁸ The Institute of Biomedical Scientists has a specific qualification with regard to advanced specimen dissection but this does not include upper GI resections. Any BMS used would therefore by trained locally, presumably under close supervision with assessment of competencies. The position of ‘Pathologists Assistant’ (PathA) has been longer established in the USA which is a similar concept to a Biomedical Scientist with higher training in dissection although the PathA may carry out further duties such as autopsy. There is no guidance in the current RCPath dataset for upper GI cancers as to who specifically should deal with these specimens.²³

Although there is no specific literature regarding the preparation of esophagogastric resection specimens, there have been studies with other tumor and tissue types, the majority of which compare sampling practices and efficacy between BMS and pathologists.

A review paper from an Italian group published in 2018 gives a good in-depth summary of literature to date on the topic and gives an overall favorable view of PathA/BMS involvement in specimen dissection. This is not only around dissection quality but also freeing up consultant time and enhancing departmental standard operating procedures.³

A Scottish study of over 1200 specimens showed that specimen dissection by a BMS was often of better quality than a consultant pathologist in terms of more rigid adherence to standard operating procedure guidelines.²⁴ These seem to have been predominantly regarding specimen description rather than sampling however; it is not clear whether final report accuracy was improved (as opposed to adherence to reporting standards). Several studies have suggested that sampling by a PathA/BMS significantly increases lymph node yield in colorectal cancer specimens as compared with pathologist sampling.^25–28 Kuijpers et al. looked at nearly 600 colorectal cancer specimen reports and found that PathAs sampled at least 10 lymph nodes in 83.2% specimens versus 60.9% of pathologist sampled specimens in one hospital (79.2% vs. 67.6% in another hospital). The PathAs also found more small (<5 mm) lymph nodes than pathologists did (64.2% vs. 53.7%).²⁷ A further study in the UK however found roughly similar numbers of lymph nodes retrieved by trained BMS staff and pathologists from colonic cancer specimens.²⁹

There is little evidence to recommend a specific person or specialist to prepare tissue blocks, but it is clear that this person should be appropriately qualified. A named lead for esophagogastric specimens should be considered to minimize variation and increase relevant case volumes.

How should lymph nodes be prepared and assessed?

In the absence of evidence, no recommendations were made.

Given the variations in practice noted in the previous HERO national practice survey⁸ with reference to preparation and assessment of the primary tumor in specimens, we considered whether recommendations could be made regarding the assessment of lymph nodes (e.g. number of levels examined). In the absence of any relevant evidence, no specific recommendations were made. Specific practices with regard to assessment of nodal tumor regression, and nodal harvest assessment more generally, are addressed elsewhere in this document.

Should adjunctive techniques for node identification be used?

In the absence of evidence, no recommendations were made.

Limited evidence exists for the use of adjunctive techniques to improve lymph node yield from gastro-intestinal specimens. A meta-analysis of predominantly colorectal, with some gastric cancer specimens, has previously reported increased nodal harvests can be achieved with the use of adjunctive techniques such as the use of fat clearing solutions, or injection of methylene blue to aid node identification.⁶ However, these studies were of mostly poor quality, and none of the described techniques are commonly used; the consensus group declined to make a recommendation on this topic.

Should minimum annual case volume targets be set for histopathologists?

In the absence of evidence, no recommendations were made.

The well-established relationship between volume and outcomes in surgery has driven many aspects of health systems policy, such as the centralization of esophagogastric surgical services.³⁰ However, whether this association holds true in pathology is less clear. Numerous studies have reported the variation in pathology outcomes such as median lymph node counts seen across different assessors.^7,17,27 There is some evidence to suggest that subspecialist pathologists will produce final diagnoses that are more predictive of outcome or may impact upon patient management following review of a non-specialist pathologist’s initial assessment.^31,32 Logically, this suggests that case volume and experience is likely to play a role. In other specialties, such as cervical cytology³³ or breast cancer,³⁴ minimal case volumes for a given assessor or assessing department have been mandated, but this is not the case in esophagogastric cancer or other gastrointestinal malignancies.

Further work is required to assess the volume–outcome relationship in pathology, and in due course may lead to the establishment of minimum case volume thresholds. Currently, insufficient evidence exists to make a recommendation in this area.

Should tissues be routinely blocked in entirety to increase lymph node harvests?

Systematic fat blocking (SFB) results in an increased nodal yield but with unclear impact on patient management or prognosis. Where adjunctive techniques to assess the specimen due to low manually dissected nodal count are considered, we recommend the use of SFB, particularly with reference to tissues attached to the lesser curve and left gastric vascular pedicle.

Strong recommendation, moderate certainty of evidence.

Evidence for this issue consists of single-center retrospective studies, which demonstrate consistently that SFB significantly results in an increased lymph node yield compared with manual nodal dissection (MND). However, additional nodes identified by SFB are more likely to be small, impalpable nodes, with no definitive evidence to suggest an impact on prognosis, staging, or post-operative management. There was no evidence stratifying different anatomical areas of nodal tissue for SFB versus MND.

Hanna et al.⁷ reported a single-surgeon case series of 114 OG resections, and examined factors affecting nodal harvest counts. While 83 specimens were manually examined, 31 underwent SFB of all peri-gastric, periesophageal, and peri-duodenal fat (greater omentum was not processed), resulting in a 32% higher mean nodal count (mean 50 vs. 66 nodes). There were significant differences between pathologists for harvest counts when using MND, this was not seen in SFB suggesting superior standardization of approach with SFB. SFB also resulted in the identification of significantly more small-diameter nodes than MND (mean size 6.4 ± 3.4 vs. 8.4 ± 1.6 mm, P = 0.023). SFB identified a greater number of tumor-containing small nodes (<5-mm diameter), though there was no statistically significant difference in number of patients upstaged through SFB or MND. Wang et al.³⁵ reported a single-center Chinese series of 725 gastrectomy specimens, comparing nodal counts for three techniques: pathologist MND, surgeon perioperative dissection of fat into separate nodal station packets with accompanying fat tissue, and surgeon perioperative dissection of individual nodes. Mean nodal harvest was 23, 31, and 46, respectively (P < 0.001). However, there was no statistical difference between the mean number of positive nodes between techniques. Similar results were reported by Afaneh et al.,³⁶ in a US series of 222 gastrectomy specimens, wherein surgeon dissection of nodal tissue packets increased median harvest from 21 to 30 nodes (P < 0.001) compared with pathologist MND. Ni Mhaolcatha et al.³⁷ compared techniques in prostatectomy specimens. Initial MND identified a median of 10 (range 1–42) nodes, with subsequent processing of all of the remaining tissue (i.e. equivalent to SFB) identifying an additional median 6 (range 0–55) nodes. This was the only study to compare pathological workload between techniques, with SFB requiring processing of twice as many tissue blocks. No evidence was found comparing SFB with other adjuncts, such as fat clearing, or repeat manual assessment; while this recommendation focuses on use of SFB, further research to compare against other methods is needed.

SFB to achieve a greater nodal harvest for surgical benchmarking or quality control must be contrasted against workload for pathologists and any implications for patient care. SFB results in greater number of tissue blocks for assessment; this increased pathology workload is only partially offset by the easier preparation of blocks compared with MND. Evidence suggests that any clinical patient benefit (i.e. through upstaging) is likely to be minimal, as most smaller impalpable nodes are benign. However, we believe that the standardization of reporting will enable surgical quality assessment and improvement which will lead to broader patients benefit as a whole. In order to promote standardization of pathological specimen analysis and surgical quality assessment, we recommend that in case of a below-threshold nodal count, that manual reassessment (see recommendation ‘Should additional assessment (and if so, what/how) be mandatory in cases where <15 nodes found?’) and/or adjunctive techniques be considered. Where adjunctive techniques are used, we recommend at a minimum, SFB of lesser curve tissues be considered in cases where initial manual assessment has failed to identify the benchmark threshold number of lymph nodes.

Should different areas of tissue and the nodes harvested from them be reported separately?

We recommend centers consider reporting of individual lymph nodes or nodal stations when there is a clinical or prognostic concern over their involvement. Individual nodes may also be separately examined or reported to document the completeness of surgical lymphadenectomy, for example common hepatic or subcarinal nodes.

Strong recommendation, very low certainty of evidence.

There is no currently available evidence to suggest the separate reporting of individual nodal stations has any impact on patient treatment. Tumor cells within different (generally, more tumor-distant) nodal stations have been shown to be associated with poorer prognosis in gastric cancer³⁸ and are the subject of the TIGER study in oesophageal cancer.²⁰ However, the pathological identification of individual nodal station status has little relevance based on current adjuvant therapeutic regimens which involve systemic chemotherapy only. We therefore recommend the selective reporting of individual station or nodes, such as in cases where clinical concern exists (e.g. to ensure successful excision in cases of extended or selective lymphadenectomy to target involved nodes) or where documentation of the extent of surgical lymphadenectomy is to be supplemented by pathological reporting of the same.

Should additional assessment (and if so, what/how) be mandatory in cases where <15 nodes found?

We recommend centers reassess specimens in cases where <15 lymph nodes are found and examined in the initial assessment. While the examination of additional nodes, which given they were not found on initial cut-up are likely to be impalpable and small in size, is less likely to have impact on patient staging, this serves as an important quality control and allows standardization for the benchmarking of surgical quality. It is anticipated that the additional workload for pathology departments from such cases is small.

Strong recommendation, very low certainty of evidence.

Although studies examining ideal lymphadenectomy thresholds to optimize survival have reported results ranging from 10 to 40 lymph nodes,^39–41 AUGIS and the US National Comprehensive Cancer Network currently recommend a minimum of 15.^42,43 In this context, this threshold acts as a surgical quality marker, with the percentage cases achieving this number used as a comparative quality indicator in the UK NOGCA. It has little impact on post-operative management, particularly given current adjuvant regimens wherein lymph node count does not affect management.

The previous HERO consensus group national survey of practice⁸ found that UK centers with the highest proportion of specimens with ≥15 lymph nodes were significantly more likely to be ones which practiced routine re-examination of specimens in the case of low initial count. This most commonly involved the use of SFB (see consensus statement relating to this, above). No other published studies which specifically examined the effect of routine specimen re-examination on reported lymph node counts were found.

If a threshold node count value is to continue to be used as a surrogate quality marker, it is crucial that thorough assessment, and re-assessment where needed, be undertaken where counts fall below this. Absolute nodal counts cannot be compared in the absence of fully standardized nodal workup (for example, routine SFB of all tissues, which the consensus group deemed unfeasible). We strongly recommend that in the interest of standardized reporting and surgical quality benchmarking, that routine reassessment of specimens not achieving the threshold nodal count be considered. As the most recent NOGCA report reported that < 15 lymph nodes were assessed in only 13.4% of esophagogastric resection specimens, the increased workload for pathology labs is anticipated to be small, though future analyses of incidence and associated cost would be of value.

Should lymphovascular invasion be routinely reported?

Although the impact of lymphovascular invasion (LVI) on treatment is greatest in early cancer, such as with T1 disease treated with endoscopic excision where LVI might weigh in the favor of radical treatment, it is an important prognostic factor across all disease stages. We recommend the routine reporting of LVI.

Strong recommendation, moderate strength of evidence.

Though not included as prognostic variables in current TNM8 staging for gastroesophageal cancer, the association between LVI and survival in esophagogastric cancer are the subject of increasing interest. Currently available evidence consists of moderate quality retrospective data with consistent outcome reporting. Recent retrospective studies have suggested LVI as an independent prognostic indicator in limited populations, such as those undergoing primary surgery without neoadjuvant therapy,⁴⁴ or in node-negative disease.⁴⁵ A 2019 meta-analysis of LVI⁴⁶ in 35 relevant studies concluded that LVI was a strong predictor of poor overall and disease-free survival, though the outdated staging and treatment protocols used in the included studies limits the generalizability of these findings. A 2020 systematic review of LVI in T1 esophageal cancer, including 23 retrospective studies (n = 4749),⁴⁷ reported that LVI was significantly associated with decreased survival, hazard ratio (HR) 1.85 (95% confidence interval 1.10–3.11) and increased risk of lymph node metastases, odds ratio 5.72 (4.38–7.4).

LVI is increasingly considered as a negative prognostic factor, particularly with reference to early (T1) disease, wherein their presence or absence may influence recommendations for radical surgery following endoscopic resection, or adjuvant treatment following surgery. We recommend its routine reporting.

How should nodal response to neoadjuvant therapy be assessed and reported?

Lymph node response following neoadjuvant therapy should be reported as it adds value to prognosis prediction and planning adjuvant therapy.

The Mandard grading system is recommended although other systems may be used. It is recommended that the results of the different systems are audited to facilitate an overall comparison to determine the most appropriate system.

Strong recommendation, moderate certainty of evidence.

The response within lymph node deposits of esophagogastric cancer to neoadjuvant therapy is becoming increasingly important as part of individual treatment planning and prognosis.^48–50 Adjuvant oncological treatments require evidence of response to determine whether to continue with the same drug combination or to change approach. This is particularly relevant as greater understanding of biomarkers increasingly helps guide treatment paradigms.

The response of tumor deposits in lymph nodes following neoadjuvant therapy has been assessed in both gastric and esophageal cancer and has included the common histological subtypes. The majority of studies show a correlation between better prognosis and increased lymph node response, independent of primary tumor staging. The definition of response remains an area lacking consensus. Studies predominantly compare pathological nodal staging or tumor regression to preoperative staging scans.^48–50 In one recent study nodal response was also assessed via changes in FDG PET-CT avidity, wherein this was a significant independent predictor of prognosis, whereas metabolic response of the primary tumor was not.⁵¹

There is variation internationally in the systems used to determine pathological nodal regression. This is largely related to inter- and intra-country variations in the use of the available systems. A recent European international survey highlighted the variation in practice, with respondents reporting use of Mandard, AJCC/Ryan, or Becker systems in 36%, 22%, or 10% of cases, respectively.⁵² Each system has a variable number of tiers and 51% surveyed preferred a four tier system. A confounding variable is that the original systems were developed from determination of regression within the primary tumor site and this has been applied to lymph node regression without full validation. The most popular Mandard system was assessed in esophageal squamous cell carcinoma and included grading of nodal response.⁵³ In addition, some studies have reported individually developed systems, these have mainly been in adenocarcinoma.^49,50,54 Reassuringly, these studies have confirmed minimal inter-observer variation whichever system is used. For purposes of standardization, we support the use of the most commonly used Mandard system. As the understanding of prognostic implications of the degree, or type, of response in primary tumor or lymph nodes evolves, the types of information of the greatest utility, and these recommendations, should change as well.

Should ratio of involved/harvested nodes be routinely reported?

In the absence of evidence, no recommendations were made.

The number of involved and harvested nodes forms part of routine histopathological reporting. Use of the lymph node ratio (involved number of nodes: total number of nodes harvested) as an explicit, separately reported entity to estimate prognosis and guide management has been recommended by some authors suggesting it is a sensitive, independent marker.⁵⁵ Its usefulness, however, is significantly limited by the non-standardized nature of reporting of harvested lymph nodes, which these guidelines hope to help address. We suggest that until the latter can be rectified, the lymph node ratio remains primarily an investigational tool at present.

Should micrometastases be routinely reported?

The yield of immunohistochemistry (IHC) to identify micrometastases is very low and of no therapeutic benefit. We do not recommend the routine analysis of lymph nodes by IHC to look for nodal micrometastases.

Strong recommendation, low strength of evidence.

Retrospective observational studies have considered the prognostic significance of nodal micrometastases detected by IHC in both adenocarcinoma and SCC. Some authors have suggested that presence of nodal micrometastases may represent a negative prognostic factor, in certain populations. Available evidence is heterogenous with almost all studies reporting sample sizes of <100 patients. A single study, by Chen et al.,⁵⁶ reported results for 516 SCC patients with cN0 disease. While this study reported micrometastases were a negative prognostic factor, potentially on par with cN1 disease, this Chinese study represents a disease and patient cohort less frequently seen in Western practice, and comprised historical patients with outdated treatment regimens (no neoadjuvant therapy).

The reported results and methods of other studies are variable. Several studies have considered only micrometastases in node-negative patients, concluding that they were a significant prognostic indicator in this cohort.^57,58 Waterman et al., conversely, found that they were not predictive in N0 patients, and that their prognostic value was limited to patients who had node positive disease with less a node positive ratio of <10%.⁵⁹ Further still, Komukai et al. reported relevance only in select populations only such as if only 4 or fewer nodes were overtly positive.⁶⁰

In contrast, a Japanese study of 50 patients with SCC by Sato et al. found that despite identifying micrometastases in 40% of patients, there was no difference in recurrence rate or survival compared with patients without micrometastases.⁶¹

On the balance of evidence, it appears probable that the presence of nodal micrometastases may predict poor outcome in patients who are N0-2 on routine pathological staging. However, the yield of micrometastases from IHC analysis is low, and typically 1–2% of overtly negative nodes examined. There is no therapeutic benefit, and the analysis is costly; we therefore do not recommend routine assessment.

What definition or format should be used for recording positive resection margins?

We recommend recording circumferential resection margins (CRM) and longitudinal resection margins (LRM) using the RCPath cancer dataset for histopathological reporting of esophageal and gastric carcinoma. This includes the use of RCPath definition of margin positivity (R1 if tumor cells at or within 1 mm of the margin), and must include the nearest distance of cancer cells from the CRM in mm in the reporting of esophagectomy specimens to allow for identification of patients with predicted worse overall prognoses.

Strong recommendation, moderate certainty of evidence.

The RCPath defines positive CRM as cases where tumor is found within 1 mm of the surgical margin,²³ whereas the College of American Pathologists (CAP) defines CRM as positive if tumor is found at the cut margin of resection.⁶²

A recent meta-analysis to examine the role of CRM status in the multimodal treatment of esophageal cancer patients included 29 studies compromising 6142 patients with heterogenous disease histology (adenocarcinoma and SCC) and neoadjuvant or adjuvant treatment regimens.⁶³ The median rate of a positive CRM was 45·3% using the RCPath definition and 17·6% according to the CAP definition. Positive CRM was predictive of poorer overall survival, independent of other disease, treatment, and patient variables, with a stronger relationship for the CAP definition, HR 2.18 (1.84–2.60), than the RCPath definition, 1.68 (1.48–1.91). Similar findings are reported in earlier meta-analyses.^64,65 Some data have suggested a relationship between the distance between tumor cells and the cut edge of the CRM.⁶⁵

The few studies reporting on LRM are limited to retrospective cohort studies. Whilst LRM has been shown to independently correlate with survival, suggested macroscopic lengths of proximal and distal margins vary from 5 to 7 cm from the primary tumor with no clear consensus.^66–68

We recommend reporting of margins in line with RCPath guidelines to ensure standardization. Within this, however, we recommend explicitly including the measured nearest distance between tumor cells and cut margin, in recognition of the likely relationship between this and outcome. The definition of margin positivity continues to be examined and is likely to be addressed in the next iteration of TNM staging.

Should the standardized RCPath reporting cancer dataset be used for reporting—does it contain sufficient data for prospective trials and standard outcome data?

The RCPath dataset for histopathological reporting of esophageal and gastric carcinoma is the minimum requirement for clinical trials and the comparison if outcome data. More detailed reporting is encouraged depending on trial requirements, but must include the cancer dataset.

Request forms, including standardized pre-treatment clinical information, and reporting of esophagectomy specimens should adhere to RCPath guidelines, but with distance of tumor cells to the nearest CRM reported in mm to allow global standardization. With improvements in accessibility to genomic data, this should be included in pathology reports where feasible.

Strong recommendation, very low certainty of evidence.

Aside from the debate surrounding reporting definitions (e.g. CRM), adherence to the UK national RCPath cancer dataset reporting guidelines has been consistently poor,^8,69 with the recent HERO national survey of practice revealing compliance with minimum reporting guidelines in only 16% of responding centers. This has important implications not only for the standardization of practice and optimization of care delivery to patients, but confounds attempts to compare outcomes between units, or conduct research to improve care. The consensus considered whether the cancer dataset was appropriate for use, or whether poor adherence was a reflection of deficiencies in the RCPath guidance. Consensus was reached with 100% of group members strongly in agreement with the existing cancer dataset.

If practice is to be improved, and collaboration between centers better facilitated, a minimum standard of practice is imperative. Standardized data throughout the specimen pathway must be aspired to. This starts with the specimen request form, which should contain all relevant clinical details such as tumor type, location, neoadjuvant treatment, surgery type, and any clinical concerns, and ends with the discussion of postoperative histology in the MDT meeting. Variations tailored to local practice, or inclusion of additional data for research purposes, are encouraged, but must be viewed as additions to the RCPath cancer dataset rather than replacements for it.

What molecular tumor markers should be routinely reported?

HER2, PDL1 and mismatch repair (MMR), or microsatellite instability (MSI) status play an increasingly important role in the management of patients in both palliative and radical treatment. Where assays are available, we recommend routine testing of diagnostic biopsies for HER2, PDL1, and MMR/MSI status on adenocarcinomas, and PDL1 in squamous cell carcinomas. If negative, consideration should be given to retesting on resection specimens including primary tumor and resected lymph nodes that contain carcinoma.

Strong recommendation, moderate quality of evidence.

(66.7% strongly agree, 11.1% agree, 22.2% disagree)

Novel molecular targeting agents to treat gastroesophageal cancer represent a rapidly evolving area of growing importance. Prospective trials have already established that patients with advanced stage gastroesophageal adenocarcinoma who are HER2 positive (defined as HER2 IHC score 3+, or IHC score 2+ and HER2 in-situ hybridization positive), benefit from the addition of trastuzumab to palliative first-line platinum fluoropyrimidine chemotherapy.^70–72

Clinically significant intra-tumoral heterogeneity for HER2 positive status in individual gastroesophageal adenocarcinoma patients has been described, including discordance between biopsies and resected specimens, leading to false-negative results and the potential for undertreatment.^73–75 Discordance of HER2 expression between primary and metastases, and between metastases at different sites has also been described.^76–78 Together this indicates the limitations for gastroesophageal adenocarcinoma treatment decisions of undertaking HER2 on only a single specimen; repeat testing in the event of a negative result should be considered when it might alter patient treatment.

Following recently reported randomized clinical trials, there is a rapidly emerging role for anti-PD1 immune checkpoint inhibitors in gastroesophageal cancer in both adjuvant and palliative settings. In gastroesophageal adenocarcinomas, HER2 negative, PDL1 positive, (defined as PDL1 combined positive score > 5) patients have been shown to benefit from addition of adjuvant nivolumab to neoadjuvant chemoradiation.⁷⁹ Preliminary analysis from randomized trials in palliative treatment of gastroesophageal adenocarcinoma indicates that MSI may be a useful predictive biomarker for pembrolizumab.^80,81 In esophageal squamous cell carcinomas, nivolumab has been shown to be superior to paclitaxel chemotherapy following progression or intolerance of first-line palliative platinum fluoropyrimidine chemotherapy.⁸²

Beyond determining oncological therapies, analysis of randomized trials indicates that resectable gastroesophageal adenocarcinomas who are MSI high, or MMR deficient potentially have a good prognosis when treated with surgery alone, leading to the suggestion that these patients could be considered for omission of (fluoropyrimidine-based) neoadjuvant therapy.⁸³

Given the rapidly evolving role for these markers, and the increasingly broad application to patient therapy, we feel that testing for HER2, PDL1, and MMR/MSI status is important both for patient care and research purposes. While some concerns were raised within the consensus group regarding feasibility and cost, the consensus opinion was that identifying tumor status for these markers will soon become the mandatory gold standard for patient treatment. Currently, it is not possible to recommend a prioritization of one test over another, should such a decision be required due to resource limitations or a lack of testable tissue; in such circumstances consideration should be given to obtaining further tissue if feasible and where it might alter patient management. While universal access to appropriate assays may not be possible at present, this reflects the rapidly evolving nature of the body of evidence for these markers; it is expected most if not all labs will have access in due course. The HERO consensus therefore recommends routine testing.

Conclusions

Current practice regarding the preparation and assessment of histopathological specimens is highly variable. We present a pan-stakeholder consensus on behalf of national surgical and pathologist bodies. These recommendations take into account the latest evidence in the management of gastroesophageal cancer, and aim to standardize practice to improve care, prognosis, and benchmarking of surgical quality.

References