Data Set for Reporting of Lung Carcinomas: Recommendations From International Collaboration on Cancer Reporting

The main objective of the ICCR is to develop an evidence-based site-specific reporting data set for each cancer by engaging pathologists with recognized international expertise in their fields to review the relevant literature and available guidelines and to use standard terminology and unambiguous descriptors and value lists for all data elements. These rules for development should be applied in a consistent manner across all of the data sets. Lastly, the information should be gathered in a way that facilitates electronic implementation by different laboratory information management systems.

To test the feasibility of data set development, a pilot project to develop data sets for prostate, endometrium, and lung cancers and malignant melanoma was set up within the quadripartite group. The liaison to ICCR from each of the 4 organizations was responsible for developing 1 of the 4 data sets (prostate: CPAC; endometrium: RCPath; lung: CAP; melanoma: RCPA). Expert panels were established for each of the cancers. Each panel was composed of 8 specialists (2 representatives from each country). Apart from 1 internationally renowned clinician from 1 of the 4 countries, all of the other representatives were specialist pathologists.

A chair was selected for each panel and was charged with the overall management of the project, directing the other panel members in a review of the 3 existing cancer protocols/data sets (RCPA, CAP, and RCPath), agreeing on a set of “core” and “noncore” data elements (subsequently redesignated as “required” and “recommended” elements; see “Comment”) for each cancer site and on permitted responses. A core (required) element was initially defined as an item of data that was essential for clinical management, staging, or prognostication of the cancer. Core data elements required supporting evidence at level III-2 or above (prognosis category, National Health and Medical Research Council Methodology)⁷  and the unanimous support of all members of the panel. A noncore (recommended) element was initially defined as an item of data that was clinically important, recommended as good practice, or regularly used for research purposes or for patient management but which had not, to date, been supported by level III-2 evidence. A generic cancer data set applicable to all sites (in the form of an Excel template [Microsoft Corporation, Redmond, Washington]) was compiled by the ICCR and provided as the starting point for each group.

As a starting point, the existing structured reporting protocols for lung carcinoma in resection specimens from the colleges of pathology in 3 of the ICCR countries, namely, CAP (American), RCPath (British), and RCPA (Australasian), were compared (for references to protocols, see the RCPA Web site: http://rcpa.edu.au/publications/structuredreporting/cancerprotocols.htm; accessed September 11, 2012). Multiple core pathologic parameters were included in the 3 national structured reporting systems and, in addition, the British and Australasian reporting protocols mandate collection of clinical history and findings and information on the surgical procedure. These additional parameters were not included in the comparison of protocols. The recommendation for inclusion of each pathologic parameter in the structured reporting protocols was classified into 1 of 3 mutually exclusive categories: concordant recommendation, minor discordant recommendation, or major discordant recommendation (Table 1). Four elements with minor discordant recommendations and 2 with major discordance were identified (Table 2).

Concordant data items did not pose a problem for the expert panel as most were required for tumor staging, grading, and prognosis. Minor and major discordant data items required careful consideration and debate, and panel members were required to provide evidence to support the categorization of the data items as core or noncore. Inclusion of core data items required level III-2 evidence. Lower levels of evidence were needed for additional (noncore) data.

Those elements considered core (required) and noncore (recommended) by the panel are considered below and summarized in Table 3 (core elements) and Table 4 (noncore) elements.

Several generic descriptors apply to all cancer types and the ICCR agreed that these should generally be included in data sets irrespective of the level of evidence in the literature to support their inclusion. Pertinent to the reporting of lung cancer, these include the type of operative procedure that generated the specimen(s), as well as several key macroscopic descriptors that include specimen/organ (lung), specimen laterality, attached anatomic structures, and accompanying specimens resected during the procedure. For the type of operative procedure, it is important to document the type/extent of resection (wedge, segmentectomy, lobectomy, bilobectomy, pneumonectomy, other) and for en bloc specimens, any attached anatomic structures received (eg, portion of rib, pericardium). Any accompanying specimens, such as mediastinal lymph nodes, must also be recorded and described. Other core macroscopic data items are detailed below.

Tumor site(s) (upper lobe, middle lobe, lower lobe, bronchus) must be documented in the pathology report. For bilobectomy or pneumonectomy specimens in which separate tumor nodules are identified (see below), the location of each tumor nodule should be clearly specified, as assignment of stage is dependent on whether nodules are within the same lobe or different lobes.

Tumor size has long been recognized as an important prognostic indicator in lung cancer.⁸  From survival data, the 7th edition of the TNM system has further subdivided the T category by tumor size.⁹  The maximum diameter of a tumor, measured to the nearest millimeter, should ideally be assessed on the unfixed specimen to avoid the possibility of size underestimation resulting from formalin fixation–induced shrinkage.¹⁰  In specimens harboring multiple synchronous primary tumors, assignment of the T category is based on the size of the largest tumor.

Care should be taken not to overestimate tumor size by including areas of adjacent obstructive pneumonia in the tumor measurement. The gross assessment of tumor size should be confirmed microscopically and in cases for which adjacent obstructive pneumonia has been mistakenly incorporated into the tumor measurement, tumor size should be adjusted accordingly.

Not infrequently, more than 1 discrete tumor nodule is identified in lung cancer resection specimens. It is important to distinguish synchronous primary tumors from a tumor displaying intrapulmonary metastases, as they have different prognoses and are staged differently.^9,11  Separate tumor nodules of different histologic types are considered synchronous primary tumors and should be recorded as such in the pathology report with the highest T category followed by the suffix “m,” indicating multiplicity, or the number of tumors in parentheses (eg, T1[m] or T1b[2]).⁹  For multiple tumor nodules with similar histologic profiles, the criteria of Martini and Melamed¹²  have long been used in this distinction. According to these criteria, tumors with similar histologic pattern are categorized as synchronous primary tumors if they are in different segments, lobes, or lungs; originate from carcinoma in situ; and there is neither carcinoma in lymphatics common to both nor extrapulmonary metastases at the time of diagnosis. More recently, comprehensive histologic assessment has been proposed as a reliable method of separation.¹³  Although a detailed discussion of this technique is beyond the scope of this article, comprehensive histologic assessment examines not only whether multiple tumors share the same major histologic pattern, but also the similarities in the percentages of other histologic patterns and cytologic and stromal features.

Patients with multiple tumor nodules deemed to not represent synchronous primary tumors in the same lobe have survival outcomes similar to those of patients with solitary tumors that by size or other criteria fall into the T3 category, and for this reason are staged similarly.⁹  Analogously, the similarity in survival between patients with multiple tumor nodules deemed to not represent synchronous primary tumors in different lobes of the same lung and patients with solitary tumors that fulfill T4 criteria has led the American Joint Committee on Cancer (AJCC) to recommend staging tumors similarly for such patients.

Assuming the margins are negative and the tumor is not of the superficially spreading type, this staging element is generally not a factor for wedge resections and lobectomies as such specimens do not incorporate the main bronchus. The proximity of tumor to the carina is a concern in pneumonectomy specimens with central tumors, particularly those that involve the right main bronchus, as it is shorter than the left main bronchus. In such cases, accurate determination of distance of tumor from the carina requires integration of clinicoradiographic data and/or consultation with the surgeon, radiologist, and/or bronchoscopist. When this information is not available, particularly as may occur in the setting of external consultation, it is permissible to indicate this staging parameter is not assessable.

The presence and extent of atelectasis/obstructive pneumonia factor into assignment of the T category. While most likely to be seen in association with central tumors that obstruct either the main or proximal lobar bronchi, this staging parameter can be difficult to accurately assess in resected specimens and often requires correlation with the radiographic findings.¹⁴  In certain instances, the lack of availability of radiologic information renders this parameter not assessable. For cases in which atelectasis/obstructive pneumonia is determined to be present, the extent to which the lung (entire lobe or entire lung) is involved should be specified.

Although level III-2 and above evidence supporting inclusion of distance of tumor to the closest resection margin as a core element is not available, the panel agreed that this information should be required to facilitate postoperative treatment planning. Documentation of the macroscopic distance between a tumor and the nearest resection margin and specification of the closest margin are invaluable in cases for which the distance is greater than that which could be encompassed in a tissue block. For cases in which the distance can be visualized on a microscopic slide, it is recommended that the macroscopic measurement be confirmed histologically.

All lung carcinomas should be typed according to the 2004 World Health Organization (WHO) classification.¹⁵  Accurate typing of lung carcinoma is becoming increasingly important, as histologic profile impacts on decisions to proceed with molecular testing (see below) and the most appropriate chemotherapy regimen for patients in whom adjuvant therapy is indicated. Given the essential role that histologic type plays in patient management, a designation of non–small cell lung carcinoma, not otherwise specified, is not acceptable in resection specimens.¹⁶  While it is beyond the scope of this article to provide a detailed discussion of the pathologic features of various histologic types of lung carcinoma, in poorly differentiated cases, immunohistochemistry can greatly aid in classification (see below).

It is anticipated that the classification of lung adenocarcinoma recently proposed by the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS)¹⁶  will in large part be incorporated in the next WHO classification, after which adenocarcinomas should be classified according to the recommended terminology.

Lung carcinomas should be adequately sampled to ensure defining features are satisfactorily represented in the sections examined histologically. For cases in which the newly proposed entities of adenocarcinoma in situ or minimally invasive adenocarcinoma are being considered, the IASLC/ATS/ERS requires that lesions be entirely submitted for histopathologic examination.¹⁶ 

It should be noted that the recommendations put forth in this document apply to small cell carcinoma and carcinoid tumors, as well as non–small cell types of lung carcinoma. While originally used primarily for non–small cell lung carcinoma, the TNM staging system has since also been scientifically validated for small cell carcinoma and carcinoid tumors.¹⁷ 

Quantification of the extent of tumor regression in patients who have received neoadjuvant chemotherapy and/or radiation therapy is prognostically useful^18,19  An estimation of whether greater or less than 10% residual viable tumor is present in the resection specimen should be reported and the “y” prefix included as part of the TNM pathologic stage.

Extension of tumor into extrapulmonary structures is an adverse prognostic factor, the degree to which depends on the structures involved.¹¹  Occasionally, lung cancer resections will include extrapulmonary structures either en bloc or separately. The presence or absence of invasion into extrapulmonary structures in such cases should be reported and the involved structures should be specified.

Lymphovascular invasion has been demonstrated to be an independent prognostic factor in lung carcinoma.^20–23  A number of studies^24–26  have evaluated the prognostic impact of large-vessel (arterial and/or venous) invasion independent of lymphatic invasion with somewhat conflicting results. For this reason, it is permissible to report the presence of vascular and/or lymphatic invasion under the single heading of lymphovascular invasion.

Completeness of resection is an important prognostic factor that also influences postoperative management, including decisions about adjuvant therapy.²⁷  The status of the surgical resection margin(s) should be reported for all resections, but the number and types of margins vary according to the specimen received. For wedge resections, the only resection margin is the parenchymal margin, which is represented by the staple line. Larger resections may include parenchymal margins (eg, lobectomies from patients with incomplete fissures) in addition to bronchial and vascular margins. Depending on the anatomy and extent of resection, these may be singular (1 bronchial margin and 1 vascular margin including an arterial and venous margin) or multiple.

A positive bronchial or vascular margin is widely considered to represent tumor within the lumen that is densely adherent to and/or involving the wall. According to several studies,^28–31  tumor restricted to the peribronchial or perivascular soft tissue at the margin, or the presence of lymphatic permeation alone at the margin, is also prognostically important. Recently, however, the significance of peribronchial soft tissue involvement without mucosal involvement has been called into question.³²  Data on the impact of intraluminal tumor alone at the margin are too limited to draw meaningful conclusions. When reporting the presence of tumor at the bronchial or vascular margin, the pathologist should provide a comment delineating the nature of the involvement.

The significance of carcinoma in situ at the bronchial margin remains unresolved owing to its rare occurrence.³³  Results of several studies^33,34  suggest that the presence of carcinoma in situ at the margin is not an independent prognostic factor. Nevertheless, it is important to report carcinoma in situ at the margin so that additional data might permit a more conclusive assessment of its role in prognosis.

En bloc resections contain additional margins (eg, rib, chest wall soft tissue), the nature of which is dependent on the type and extent of extrapulmonary structures resected. Ideally, the surgeon will designate the location of the resection margin(s) of extrapulmonary structures before submission of the specimen, but in ambiguous cases, direct communication will help to ensure appropriate handling and submission of tissue for histopathologic examination. The status of additional margin(s) and their location(s) should be specified in the pathology report.

Lymph node metastases are an adverse prognostic factor, the extent to which is dependent on the location of the involved lymph nodes.³⁵  The lymph node status should be reported as the number of lymph nodes involved and the total number of lymph nodes submitted, specifying the site(s) of involvement (lymph node stations) according to the IASLC lymph node map.¹¹  Given the nature of the procedure, lymph nodes obtained by mediastinoscopy are often received fragmented and unless specified by the surgeon, it may not be possible to distinguish a single fragmented lymph node from fragments of multiple lymph nodes. For this reason, when a determination of the actual number of nodes is not possible, it is permissible to report the sites of nodal metastases without specifying the number involved.

The presence of tumor at the surface of the visceral pleura has been recognized as an independent adverse prognostic factor for quite some time.⁸  More recently, penetration through the visceral pleural elastic layer was shown to have the same prognostic impact.^36,37  With the release of the current staging classification, criteria for visceral pleural invasion (VPI) have been more clearly defined to encompass both invasion beyond the visceral pleural elastic layer and extension to the visceral pleural surface.³⁸  For tumors that are in contact with the visceral pleura and do not clearly extend to the visceral pleural surface, elastic stains can aid in the detection of tumor cells beyond the visceral pleural elastic layer.

Oftentimes, there is not 1, but 2, perceptible visceral pleural elastic layers. In most individuals, the elastic layer that is closer to the surface of the visceral pleura, typically referred to as the outer or external elastic layer, is thicker and more continuous, while within the visceral pleural connective tissue adjacent to the alveolar parenchyma lies a less prominent and/or somewhat fragmented internal (inner) elastic layer. It is the recommendation of the International Staging Committee³⁸  that the thickest elastic layer be used to assess VPI. Occasionally, tumor cells are intermingled with fibers of the visceral pleural elastic layer without unequivocally penetrating beyond the visceral pleural elastic layer. This should not be interpreted as evidence of VPI.

A small percentage of cases are indeterminate for VPI. Occasionally, the visceral pleural elastic layer is imperceptible, even with elastic stains, in cases for which tumor is in contact with the visceral pleura but does not extend to the visceral pleural surface. In such circumstances, the TNM classification dictates that the lower category be assigned (ie, tumors should not be upstaged on the basis of equivocal VPI).¹¹  So too is the case for which the visceral pleura in the vicinity of a tumor is fibrotic or elastotic to the point of obscuring the normal visceral pleural elastic landmarks, such that elastin staining is difficult if not impossible to interpret. Rarely, owing to adhesions or other technical factors, a specimen is received devoid of visceral pleura overlying a tumor and it is simply not possible to assess VPI.

Data on tumors that cross an interlobar fissure but are not present on the visceral pleural surface are limited, but under current staging recommendations, these tumors are categorized as T2.³⁸ 

The pathologic TNM (pTNM) stage should be documented in the pathology report. Occasionally, the pathologic stage is higher than that which would be apparent from the specimen being examined (eg, an antecedent or separately submitted pleural fluid demonstrating malignant cells). Such information, if known to the pathologist, should be incorporated into assignment of the pTNM with a comment detailing how the pathologic stage was derived.

Although a tiered grading scheme for lung cancer is specified by the AJCC, its reproducibility and prognostic significance have not been rigorously tested.³⁹  Recently, a system of grading tumors, based on histologic pattern, has shown that tumors can be separated into prognostically distinct groups.⁴⁰  Validation of this proposed system will require additional studies.

The use of immunohistochemistry to aid in the histologic typing of lung carcinoma is a broad topic, a comprehensive discussion of which is beyond the scope of this article. A few points will however be emphasized.

As discussed in a preceding section, a concerted effort should be made to classify poorly differentiated lung cancers in resection specimens. A few studies^41,42  have examined the best means for doing so by using an immunohistochemical approach, which has shown thyroid transcription factor 1 (TTF-1), napsin, cytokeratin 5/6 (CK5/6), and p63 to be among the most reliable markers. Moreover, p40, an antibody against an isoform of p63, has recently been reported to be a highly specific marker for squamous cell carcinoma.⁴³ 

Mucinous adenocarcinomas of the lung can exhibit aberrant staining for markers that are more commonly associated with carcinomas of the gastrointestinal tract, such as CK20 and CDX-2, and/or fail to stain with markers typically associated with pulmonary carcinoma, such as CK7 and TTF-1.⁴⁴  In such cases, exclusion of metastasis from an extrapulmonary primary tumor is best achieved by careful correlation with the radiographic distribution of disease.

A detailed review of the rapidly growing number of publications on the significance of epidermal growth factor receptor (EGFR) mutations and other molecular alterations in lung cancer falls outside the scope of this article. Aspects pertinent to the reporting of lung cancer resection specimens will be highlighted.

A small proportion of lung adenocarcinomas harbor mutations in the epidermal growth factor receptor (EGFR) gene that make them susceptible to the EGFR tyrosine kinase inhibitors (EGFR-TKIs) erlotinib and gefitinib.^45,46  Originally reported to occur most frequently in young female East Asian never-smokers whose tumors had a prominent lepidic (designated at the time as bronchioloalveolar) growth pattern, TKI-responsive EGFR mutations have also been documented in patients with other demographic and clinicopathologic characteristics.¹⁶  EGFR-TKIs improve progression-free survival for patients who have lung adenocarcinoma with EGFR mutations, to the point that these agents are now indicated as first-line therapy in advanced-stage disease.⁴⁷  For this reason, the IASLC/ATS/ERS has recommended that patients with advanced-stage lung adenocarcinoma have their tumors tested for the presence of EGFR mutations, with DNA sequencing as the preferred method of analysis.¹⁶  It is anticipated that forthcoming guidelines jointly proposed by the CAP, the IASLC, and the Association for Molecular Pathology will expand the recommendation for EGFR mutational testing to include all lung adenocarcinomas.⁴⁸ 

KRAS mutations; expression of excision repair complementing factor 1 (ERCC1), ribonucleotide reductase M1 (RRM1), and thymidylate synthase (TS); and EML4-ALK translocations are but a few of the continuously expanding array of molecular alterations other than EGFR that have prognostic and/or therapeutic implications in lung cancer.

Mutations in KRAS are associated with a lack of response to EGFR-TKIs.⁴⁹  High expression of the enzyme ERCC1 predicts resistance to platinum therapy and shorter survival.^50,51  Low expression of RRM1 is associated with improved survival with gemcitabine/platin therapy.⁵⁰  High expression of TS confers a less favorable response to a class of drugs that includes 5-fluorouracil.⁵²  At present, testing for these molecular alterations is at the discretion of the reporting institution and/or preference of the treating physician.

EML4-ALK translocations, like EGFR mutations, occur in a small subset of patients with lung cancer, most typically never-smokers or light smokers with pulmonary adenocarcinoma, and are the target of a selective chemotherapeutic agent.⁵³  The recently discovered drug crizotinib significantly improves progression-free survival for patients with lung carcinoma harboring EML4-ALK-translocations.⁵⁴  EML4-ALK translocations are nearly always mutually exclusive of EGFR and KRAS mutations.⁵⁵  Given the efficacy of crizotinib, it appears likely that testing for EML4-ALK translocations in lung adenocarcinomas that lack EGFR and KRAS mutations will become standard of care in the near future. The National Comprehensive Cancer Network⁵⁶  has in fact recommended that patients with advanced-stage nonsquamous non–small cell carcinoma be tested not only for EGFR mutations, but also for ALK translocations. The preferred and only US Food and Drug Administration–approved method for EML4-ALK translocation testing is a fluorescence in situ hybridization assay that uses a break-apart probe.⁵⁷  Studies of other detection techniques, such as using an immunohistochemical marker that is specific for EML4-ALK, are ongoing.⁵⁸ 

Provided they are not present at the margins, the presence of other neoplastic processes (eg, tumorlets, minute meningothelial nodules, neuroendocrine hyperplasia, atypical adenomatous hyperplasia, squamous dysplasia) does not impact tumor staging. Nevertheless, the reporting of other neoplastic processes can be useful in that it may provide correlation with radiographic abnormalities not directly related to the tumor.

The reporting of nonneoplastic pathologic pulmonary findings may also provide useful correlative information to radiographic abnormalities not directly related to the tumor. The recognition and reporting of certain findings, such as usual interstitial pneumonia or pneumoconioses, can have significant impact on patient management and/or prognosis.

The macroscopic appearance of the visceral pleural overlying a tumor can help to guide the submission of tissue blocks and gauge the index of suspicion for VPI. It is important to note, however, that macroscopic visceral pleural puckering is not itself diagnostic of VPI.³⁸  The presence of VPI must be confirmed histologically.

Although tumor penetration beyond the visceral pleural elastic layer has been shown to have the same prognostic significance as tumor extending to the visceral pleural surface (see above), the pathologist may wish to provide greater detail in the report by documenting the extent of pleural invasion. A scheme for classifying pleural involvement by tumor, put forth by Dail and Hammar⁵⁹  and which has been recognized by the Japan Lung Society and recently undergone slight modification by the International Staging Committee,³⁸  is as follows: PL0, no penetration beyond the visceral pleural elastic layer; PL1, tumor penetration beyond the visceral pleural elastic layer; PL2, tumor extension to the visceral pleural surface; and PL3, extension into the parietal pleura. PL0 is categorized as VPI absent, while both PL1 and PL2 types of VPI change the category of otherwise-T1 tumors to T2. Tumors that would otherwise be categorized as T1 or T2 are changed to T3 in the presence of type PL3 pleural involvement.³⁸ 

The nature and site of all tissue blocks submitted for histopathologic examination should be recorded, ideally in the final pathology report. This will considerably facilitate accurate internal or external review/consultation. If this information is not included in the final pathology report, it should be readily accessible in the laboratory information system, such that it can be expeditiously provided to a reviewing pathologist.

The prognostic impact of perineural invasion has been investigated in several studies,^25,60–62  the results of which have been mixed. Perineural invasion has therefore been designated as a noncore element and its reporting is at the discretion of the pathologist.

The various expert tumor panels (prostate, endometrium, lung, malignant melanoma) were largely successful in achieving the main objectives of the ICCR, namely, the development of evidence-based reporting data sets that used standard terminology and descriptors. A report from the ICCR Endometrial Carcinoma Working Group is being published in the International Journal of Gynecologic Pathology.⁶³  Each panel was able to agree on “core” and “noncore” data elements and permitted responses for the selected cancer sites within the agreed time frame of approximately 6 months. The lung expert panel was able to exclude outdated and unvalidated data items and thereby simplify and improve upon the content of the existing data sets. The main challenges for all the panels related to communication, the specific method of data set development, and the variability of the permitted responses (value lists) across the data sets.

The lung panel communicated primarily by e-mail. Because of the widely disparate time zones, conference calls and Webinars were not used because of the difficulty in scheduling, although these methods may improve the efficiency of the process. For more contentious cancer checklists, at least one face-to-face meeting, possibly linked to an international pathology meeting, might help to resolve some of the more problematic issues, although such a meeting was not needed to generate the lung minimum data set.

The support of a project manager, who developed an initial suite of data elements complete with supporting evidence and responses for the panel to consider, greatly facilitated the development of the melanoma data set. In the lung panel, consensus was achieved through a more drawn-out iterative process whereby panel members provided individual responses to each of the core and noncore data items with supporting references, and the cochairs had the responsibility of moderating and collating all the responses. This was more labor-intensive and time-consuming for chairs and other panel members alike. The ICCR has therefore recommended that the process (and the final design and structure of future collaborative data sets) should follow the model used to develop the melanoma data set. Unlike the prostate, endometrial, and lung data sets, which were produced as Excel spreadsheets, the project manager of the melanoma data sets produced a detailed version of the cancer data set as Word (Microsoft Corporation, Redmond, Washington) and PDF documents, which was easier to follow and which incorporated all relevant citations and commentary as hyperlinks for ease of use. The endometrial cancer data set has since been converted to this format and is available at http://www.rcpa.edu.au/Publications/StructuredReporting/ICCR.htm (accessed September 10, 2012).

Several members of the expert panels expressed reservations about the use of the words core and noncore to describe data items and felt that the meaning of these words was not clear in the context of cancer data sets and did not convey the status of the data items accurately. The ICCR therefore agreed to use the terms required and recommended instead, and clarified the definition of these terms. Required data items were redefined as those items essential for the diagnosis, staging, assessment of prognosis, and clinical management of a given cancer. Recommended elements fell into 2 groups: (1) those data items that were regularly used for research purposes, patient management, or epidemiology but were not supported by level III-2 evidence, and (2) data items that were thought to be essential for good practice, such as the macroscopic descriptors and block designation.

Data set development in isolation by small groups of specialist pathologists in separate countries, without user feedback, tends to favor the incorporation of elements that are of particular interest to the individual pathologists who are responsible for writing the data sets. Such data sets may include unnecessary detail and mandate the recording of unvalidated elements of questionable clinical utility. Individual pathologists may find it difficult to accept and adopt change when data sets have been in use for some time, but dispassionate review of data set items by an international expert group, whose members were ready to challenge dogmas and to require justification for individual data items, led to the development of simpler, more focused data sets.

The ICCR aims to expand the process of international cancer data set development beyond the initial quadripartite alliance. In 2012, the project will extend beyond the original partners to include non-Anglophone countries in the development of an international data set for renal cancer through collaboration with the International Society for Urologic Pathology. In the short term, the 4 data sets that have already been developed will be subject to international consultation, review, and validation. In the longer term, the ICCR seeks to partner with an international cancer organization to facilitate the production and adoption of internationally developed, validated, evidence-based data sets for the reporting of all cancers. Through the use of such data sets by the worldwide pathology community, not only will the quality and consistency of cancer reporting improve but also the availability of improved cancer data should facilitate research, international benchmarking, and epidemiologic comparisons, and make for more efficient use of health care resources.