Diagnosing lung cancer in the 21 st century : are we ready to meet the challenge of individualized care ?

Background Histologic and molecular subtyping have become increasingly important as predictors of treatment benefit in lung cancer. The objective of the present study was to determine whether current diagnostic approaches provide adequate tissue to allow for individualized treatment decisions. Methods Our retrospective cohort study of new lung cancer patients seen at an academic centre between July 2007 and June 2008 collected baseline demographic and diagnostic information, including mode of diagnosis, type of diagnostic material, and pathology diagnosis. Results Of the 431 study patients, 20% had stage i or ii non-small-cell lung cancer (nsclc), 24% stage iii disease, and 39% stage iv nsclc. Three quarters of the small-cell lung cancer (sclc) cases were extensive stage. Diagnostically, 18% of patients had sclc; 30%, adenocarcinoma; 27%, squamous-cell cancer; 2%, large-cell carcinoma; 1%, bronchoalveolar carcinoma; 1%, mixed histology; 18%, nsclc not otherwise specified; 4%, other; and 2%, no pathology diagnosis. Surgical pathology material was available in 80% of cases, and cytology material alone in 20%. Surgical pathology material was more common in patients with early-stage than with advanced disease (89% for stages i and ii vs. 74% for stages iii and iv, p < 0.0001). The pathology report included ambiguous terms in 24% of cases: “consistent” (12%), “suspicious” (3%), “favour” (2%), “suggestive” (2%), “likely” (1%), “compatible” with malignancy (1%), “at least” (1%), “atypical” (0.5%), and “no pathology” (1.5%). Conclusions Current diagnostic approaches in most lung cancer patients appear adequate, but complete histopathologic identification is missing in nearly 20% of cases, and some uncertainty as to the final diagnosis is expressed in 24% of pathology reports. Some improvement in diagnostic sampling and pathology reporting are required to allow for implementation of current treatment approaches.


INTRODUCTION
Lung cancer is the most common cause of cancer death in Canada and therefore represents a major health issue.In 2012, there were 25,600 new lung cancer cases and 20,200 deaths from lung cancer 1 .Long-term survival is poor primarily because most patients present with either locally advanced or metastatic disease.Modest gains in survival have been made since a Non-small Cell Lung Cancer Collaborative Group meta-analysis, published in 1995, established first-line platinum-based chemotherapy as a standard of care 2 .Since then, several cytotoxic agents have been introduced into first-and second-line treatment, with modest survival gains [3][4][5][6][7] .
Historically, the histologic subtype of non-small-cell lung cancer (nsclc) has not influenced chemotherapy treatment algorithms 8 .Initial trials of erlotinib, an epidermal growth factor receptor (egfr) tyrosine kinase inhibitor, demonstrated benefit across a spectrum of nsclc patients 9 .Subgroup analyses suggested a benefit from erlotinib in all patient subgroups, and that agent was therefore incorporated into treatment algorithms applicable to all nsclc patients 10 .
An additional molecular target identified in lung cancer is vascular endothelial growth factor.The addition of bevacizumab, a monoclonal antibody against circulating vascular endothelial growth factor, to carboplatin and paclitaxel chemotherapy demonstrated improved survival in patients with advanced nsclc 11,12 .However, there appeared to be an association between histology and toxicity.Squamous histology is a relative contraindication to therapy with bevacizumab, because of concerns about fatal pulmonary hemorrhage 13 .
In recent years, further evidence emerged emphasizing the importance of histology as a predictive factor for systemic treatment.Histology appears important in selecting patients likely to benefit from pemetrexed 14 .In vitro data show greater expression of thymidylate synthetase, the primary target of pemetrexed, in squamous cancers 15 .Those findings suggest a potential resistance mechanism to pemetrexed chemotherapy in such patients.Clinical data support those observations.Retrospective data from a randomized trial of second-line pemetrexed versus docetaxel demonstrated a significant treatment interaction by histology 16 .Among patients with nonsquamous histology, those randomized to pemetrexed (compared with those randomized to docetaxel) experienced improved survival.The reverse effect was observed in patients with squamous histology.Similar findings were demonstrated in prospective evaluations of two additional trials comparing first-line cisplatin-pemetrexed with cisplatin-gemcitabine and maintenance pemetrexed after 4 cycles of a platinum doublet 17,18 .The trial by Ciuleanu and colleagues 17 showed no benefit from maintenance pemetrexed in patients with squamous histology, suggesting that such patients should not receive therapy with pemetrexed.Histology is now considered to be an important factor in predicting both efficacy and toxicity from newer treatment options.
The results of molecular analyses of tumours also appear predictive of treatment benefit for patients with lung cancer.Data from the i-pass trial, comparing gefitinib with carboplatin and paclitaxel as first-line therapy for advanced nsclc, demonstrated a significant improvement in progression-free survival favouring gefitinib for patients with activating mutations of the EGFR gene 19 .However, patients known to be wild-type EGFR experienced inferior progression-free survival with first-line gefitinib therapy.In total, seven trials have demonstrated the predictive value of EGFR mutation status and initial therapy with an egfr tyrosine kinase inhibitor [19][20][21][22][23][24][25] .The presence of a translocation of the ALK gene also appears to predict treatment benefit.A phase iii trial in ALK-positive nsclc patients that compared crizotinib, an Alk inhibitor, with second-line chemotherapy demonstrated improved progression-free survival for patients randomized to crizotinib 26 .Additionally, there appears to be a small improvement in survival for patients treated with cetuximab, a monoclonal antibody against egfr, in combination with platinum-based chemotherapy 27 .Overexpression of egfr protein appears to be predictive of benefit from cetuximab.
It is clear that the management of nsclc has become increasingly complex, and treatment algorithms for each patient depend on both histology and molecular analysis.Historically, the pathologist had only to distinguish between small-cell lung cancer and nsclc to enable appropriate treatment decisions.However, pathology examination of diagnostic material has become an integral component of individualized treatment decision-making for patients with lung cancer.
The diagnosis of lung cancer is frequently made on small diagnostic samples.The proportion of patients who undergo surgical resection of lung cancer and have large amounts of tumour available for analysis is only about 20%.For the remaining patients, the amount of diagnostic tissue is often small and might not be amenable to immunohistochemical (ihc) testing and molecular analysis.It is therefore unclear whether current approaches to the diagnosis of lung cancer will provide adequate pathology samples to allow for individualized treatment decisions to be made.
The objective of the present study was to examine the quality of pathology reports for patients with lung cancer to determine the suitability of those reports for making individualized treatment decisions using newer treatment algorithms, and to determine whether existing diagnostic approaches provide adequate tissue for pathology assessment and molecular analysis.

METHODS
Patients with a new lung cancer diagnosis presenting to the Juravinski Cancer Centre (jcc), an academic centre, during a 1-year period were retrospectively reviewed.All newly referred patients seen between 1 July 2007 and 30 June 2008 were eligible for inclusion in the review.Patients treated at another institution and referred to the jcc specifically for radiation were excluded, because details of their systemic therapy were generally not available.The study was approved by the Hamilton Health Sciences Research Ethics Board.
Patients were identified using the cancer centre's electronic database, and their charts were reviewed.Patient demographics, stage of disease, diagnostic tests performed, and treatment for lung cancer were recorded.Pathology reports were identified from the cancer centre patient chart or the electronic hospital chart system (or both).If the pathology report was not available in either location, then a copy of relevant pathology reports were requested from the original institution.If more than one report was available, then data from the most definitive pathology sample was used [the hierarchy used was surgical resection, mediastinoscopy, core biopsy, fine-needle aspiration (fna) biopsy, cytology].Data relating to the tumour sample, including source of the biopsy and the mode of diagnosis (bronchoscopy, needle biopsy, mediastinoscopy, surgical resection, or other), were abstracted from patient records.The type of material, either histology or cytology, was also recorded.The pathology diagnosis and any uncertainty in the wording about malignancy were collected."Uncertainty" was defined as the use of words such as "suspicious," "likely," "consistent," "favour," or "suggestive" in association with the diagnosis of malignancy.Each patient's current vital status was ascertained from the chart.If the patient's status was unknown or if the patient had not been seen at the jcc in the preceding 3 months, the family doctor was contacted to determine status, including date of death where applicable.
Data were abstracted onto data-entry sheets.The IBM SPSS Statistics software application (IBM, Armonk, NY, USA) was then used to enter the data into an electronic database.The primary outcome of interest was the proportion of patients for whom sufficient information about histologic subtype was available to make appropriate treatment decisions.Secondary outcomes included the proportion of patients diagnosed using cytology compared with histology, the proportion of patients with sufficient material to permit molecular testing, and the proportion of patients for whom there was some uncertainty in the pathology diagnosis.
The data analyses were largely descriptive and exploratory, and formal sample size calculations were therefore not undertaken.Actuarial survival was calculated using the Kaplan-Meier method, and exploratory analyses using a log-rank test were undertaken to examine survival as a function of the type of diagnostic material, either histology or cytology.The study period of 1 year was chosen to ensure that the sample of patients was representative and that the patient numbers were sufficient to include those who might be candidates for second-or third-line therapies, for which several decisions depend on histologic or molecular testing.

RESULTS
During the study period, 431 new lung cancer patients were seen at the jcc.Table i summarizes patient characteristics.Median age was 69 ± 10 years in the group, which consisted of slightly more men (55%) than women (45%).Most patients were current or former smokers, and most (80%) had nsclc.Table i also summarizes information on disease stage.A large proportion of patients presented with either locally advanced or metastatic lung cancer.
Patients in the study cohort underwent a variety of diagnostic tests to establish their diagnosis of lung cancer (Table ii).A large proportion (61%) underwent bronchoscopy.Core biopsies were performed in 34%, and 22% underwent surgical resection.In approximately 13% of patients, diagnostic material was obtained from areas other than the chest.Results for the most definitive method of diagnosis differed somewhat.Bronchoscopy was the primary diagnostic test about one third of the patients (32%).In 28%, the primary method of diagnosis was a core biopsy, and it was surgical resection in 22%.The diagnosis was also made by mediastinoscopy (3.5%), needle biopsy (6%), other means (6%), and imaging alone (2%).
Cytologic diagnostic material was available for 47% of the patients, and it was the most definitive diagnostic material in 19%.Histologic or surgical pathology samples were available for 79.4% of the patients.The diagnosis of lung cancer was made on imaging studies alone in 1.6% of cases.For most patients (83%), the origination site of the diagnostic material was lung."consistent" (12%), "suspicious" (3%), "favour" (2%), and "suggestive" (1.6%), as well as "compatible" (0.7%), "likely" (0.7%), and "at least" (0.7%) were frequently used to describe the diagnosis of cancer.Use of uncertain terminology was more frequent for the cytology samples than for the histology samples (59% vs. 13%, p < 0.0001).An exploratory analysis examined the association between the type of pathologic material and overall survival.Median survival in the entire cohort was 8 months.Survival was significantly better in patients with histology samples than in patients with cytology samples (9 months vs. 6 months, p = 0.001).Because that observation might have been confounded by stage, the analysis was repeated for the patients with stages iiib and iv disease only.The results were similar, showing median survivals of 7 months and 4 months respectively (p = 0.037).

DISCUSSION
In this comprehensive review of pathology reporting for patients with newly diagnosed lung cancer, most patients appear to have diagnostic samples adequate for establishing the diagnosis.However, our data suggest that, for as many as 1 in 5 lung cancer patients, information to make appropriate treatment decisions might be insufficient.The findings highlight two key barriers to the implementation of individualized treatment decisions for patients with lung cancer: n A subset of patients have insufficient diagnostic material.n Information from pathology reports on small biopsy samples is often incomplete.
Our data demonstrate that nearly 20% of lung cancer diagnoses are made using cytology alone.In light of changing practice patterns, that subset of patients might not have enough diagnostic material available for individualized treatment decisions that require molecular testing.The information in the pathology reports was insufficient to permit comment on the adequacy of the samples for ihc or molecular testing.In general, however, cytology samples are frequently inadequate for such testing.The specific treatments available for patients with nsclc harbouring activating mutations of EGFR or translocations of ALK necessitate routine testing for those molecular abnormalities in all patients with advanced nonsquamous nsclc.Our data suggest that up to 20% of patients would not have samples appropriate for such testing.Preliminary clinical data concerning the effects of additional molecular abnormalities such as KRAS 28 and ROS1 29 mean that the profile of the molecular tests required in lung cancer diagnostic samples will only expand as additional molecularly targeted therapies emerge.
Our data raise questions about the current approach to the diagnosis of lung cancer.In our study, 32% of patients underwent bronchoscopy, and 6%, a fna biopsy as the most definitive mode of diagnosis.Bronchoscopy can result in a surgical pathology sample for patients having central tumours with an endobronchial component.More often, though, bronchoscopy results in washings or brushings that provide minimal diagnostic samples that are frequently insufficient for ihc testing and molecular analysis.Those findings question the role of bronchoscopic brushings and washings for the definitive diagnosis of lung cancer.Additionally, the use of fna biopsy for the diagnosis of lung cancer needs to routinely incorporate the preparation of cell blocks to maximize the yield of tissue for subsequent testing.
Small diagnostic samples also create issues with respect to pathology reporting.Histologic subtyping was not available for 23% of the diagnostic samples in the present study.Treatment decisions for those patients can thus be limited, because the use of agents such as pemetrexed and bevacizumab in systemic therapy has been shown to be ineffective-or even harmful-in patients with squamous histology.For most of those patients, the pathology report gave a diagnosis of nsclc nos.Some in

Report term Use [n (%)]
Malignant 329 ( this group of likely heterogeneous patients have poorly differentiated tumours for which further classification is not possible, but in others, subclassification would be possible with additional ihc testing for markers of squamous or glandular differentiation 30 .Recommendations in a recently published systematic review suggest that histology can be prognostic, but is clearly predictive of treatment efficacy or toxicity 31 .Every effort should be made to establish a diagnosis of either squamous or nonsquamous malignancy in all nsclc samples.In the present study, it is unclear whether insufficient material for testing or a lack of awareness among pathologists contributed to the results.The issue of subclassification in nsclc has previously been studied.The proportions of subclassifications in our series of patients are similar to those in the reported literature.A recent study examined 602 fna samples and found that although 25%-35% of samples were initially categorized as nsclc nos, only 17% could not be subclassified after further ihc review 32 .Another small study found that, after pathology review of cytology specimens initially diagnosed as nsclc nos, only 12.5% of specimens still could not be subclassified 33 .Educating pathologists about how these issues influence treatment decisions is key to implementing change.
An exploratory analysis examining survival in patients with stages iiib and iv disease raised the possibility that patients with inadequate diagnostic material might experience worse survival.Although that analysis is subject to multiple potential confounders, it does emphasize the potential for diagnostic approaches to influence patient outcomes.Ambiguity in the pathology diagnosis might result in selection of less-than-optimal treatment and emphasizes the need for acquiring appropriate pathology samples in all patients with lung cancer.
The challenges inherent to the use of small diagnostic samples for lung cancer have been raised in the patholog y, respirolog y, and oncolog y literature [34][35][36] .Canadian consensus guidelines and recommendations from the International Association for the Study of Lung Cancer, the American Thoracic Society, and the European Respiratory Society all emphasize the need for a sufficient quantity of diagnostic material to allow for ihc and molecular testing, and highlight the need for a new approach to small biopsy samples 31,37 .Although the issue of small tumour samples in advanced lung cancer has become a topic of interest in the literature, the proportion of patients diagnosed using cytology alone rather than histology has not previously been reported.Earlier studies examined the question of whether cytology is sufficient for molecular analysis and have concluded that, although such an analysis might sometimes be possible in selected patients and with the use of newer pathology techniques, there are still patients for whom cytology is insufficient [38][39][40] .Furthermore, the use of ihc testing to subclassify nsclc requires tissue and can further deplete the material available for molecular analysis 41 .
An important issue for consideration is how to implement a change in practice.There is a need for knowledge translation approaches such as audit and feedback to provide information from this study both to physicians involved in obtaining diagnostic samples in lung cancer patients (respirologists, thoracic surgeons, and interventional radiologists) and to pathologists reporting on the samples.Regional educational meetings with physicians who are involved in the diagnosis and reporting of lung cancer are needed to provide feedback on our findings and to discuss strategies for implementing change.The sessions should have a knowledge component emphasizing the evidence driving the change in practice and should highlight gaps in existing pathology reporting for small biopsy samples.
An additional strategy for consideration is the standardization of pathology reporting for small tissue samples.Routine preparation of a cell block and ihc testing on all samples that cannot be classified morphologically should be considered.Additional information could be added to the pathology report, including whether a cell block was prepared from the sample and whether the sample is adequate for molecular testing such that for as EGFR mutation and ALK translocation.
Lastly, elimination of the use of ambiguous terminology and, in cases in which diagnostic material is insufficient, inclusion of specific comments recommending the collection of additional diagnostic material (if possible) could potentially improve the quality of pathology reporting.

CONCLUSIONS
The findings presented here represent 1 year of new lung cancer patient referrals to a single academic centre, a focus that might limit the generalizability of the associated findings.However, it is likely that other institutions across Canada and elsewhere are facing similar challenges with respect to the diagnosis of lung cancer and the reporting of pathology samples.Those challenges potentially require a paradigm change in the approach to the diagnosis of lung cancer, with a greater focus on obtaining histologic material whenever possible.Inherent in our recommendations are clinical challenges such as the risk of more invasive procedures, increased technical difficulty in some cases, and a potential delay in the start of treatment while waiting for diagnostic testing.However, adopting personalized care for patients with lung cancer requires that we facilitate discussions with our colleagues in respirology, thoracic surgery, and pathology about the importance of obtaining adequate diagnostic material and making improvements in pathology reporting for patients with lung cancer.
Table iii summarizes the pathology diagnoses.Most patients had nsclc.Similar proportions of patients had adenocarcinoma (29%) and squamous cancer (27%).In 23% of the patients, the pathology diagnosis would have been inadequate for treatment decision-making [nsclc not otherwise specified (nos): 18%; no pathology diagnosis: 1.9%; other: 3%].Patients with small-cell lung cancer constituted 17% of the cohort.Significant uncertainty was evident in the wording of 22% of the pathology reports (Table iv).Words such as

TABLE I
Demographics of the study patients

TABLE III
Summary of pathology diagnoses