Case 4 -

Metastatic versus Primary Carcinoma of the Lung Dr. Keith Kerr Department of Pathology, Aberdeen University Medical School and Aberdeen Royal Infirmary Scotland

This presentation considers a case of malignant epithelial tumour of the lung that was originally diagnosed as a primary tumour but subsequently proved to be a solitary metastasis from an extra-thoracic lesion diagnosed and treated many years before. The case is illustrative of the problems faced by pathologists in dealing with the issue of whether a tumour in the lung is primary or metastatic. This short paper reviews some background data on pulmonary metastatic disease and considers the diagnostic challenges of distinguishing primary and secondary carcinoma in the lung, predominantly in the context of a solitary mass lesion. Clinical History: A 46 year old male Caucasian was referred to his local district general hospital with a history of cough productive of purulent sputum and pain over the right side of his chest. He was a life-long cigarette smoker. On examination he had finger clubbing. Chest CT scan showed a 7cm right lower hilar mass apparently indenting the right atrium. A 1.8 cm nodule was also noted in the left lower lobe. Extensive mediastinal lymphadenopathy was reported. Bronchoscopy, including biopsy and cytology was negative but percutaneous trans-thoracic fine needle aspiration cytology of the left lower lobe mass shower unequivocal malignancy reported as squamous cell carcinoma. He was deemed inoperable and given chemotherapy with apparent response. Five years later he re-presented to hospital after being lost to follow-up with recurrence of right-sided chest pain. Chest CT again showed the right hilar mass lesion, the left lower lobe lesion measured 1.2 cm and there was no evidence of mediastinal lymphadenopathy. He was reviewed at the regional cardiothoracic surgery centre. Detailed work-up showed no evidence of disease outside the lungs and the patient underwent wedge biopsy of the left lower lobe nodule and later, right pneumonectomy. The tumour was adherent to the pericardium and oesophagus but there was no mediastinal lymphadenopathy. Postoperatively he received radiotherapy and had over three symptom-free years before death from myocardial infarction 3 years after surgery. At necropsy recurrent tumour was evident in the mediastinum. Case 4 - Slide 1 Click to view with ImageScope Click to view with a Web-Based Viewer Background [1, 2, 3] Primary carcinoma of the lung is the commonest fatal malignant disease worldwide, yet metastatic disease to the lungs from other primary sites is even more commonly encountered in clinical practice. The reasons for this are numerous but include the unique position of the lungs, unlike any other organ in the body, in receiving the entire cardiac output. The mixed venous blood which enters the pulmonary circulation has only recently mixed with lymph drained from much of the body via the thoracic duct and thus one of the densest capillary networks in the body is the first to encounter any circulating tumour cells. Add to this the relatively high oxygen tension in pulmonary tissue and other unknown factors that make the lungs favourable 'soil' for the seeding of metastases. The haematogenous route is thus the most important way in which metastasizing tumour cells reach the lung, but (retrograde) lymphatic spread is also important. Tumours may metastasize via the airspaces though this is likely to be an exceptional route for an extra-pulmonary tumour to reach the lung. Autopsy studies show metastases to the lungs in around half of cancer deaths; only the liver has more. In 15 – 25% of cases the lungs are the only sites of metastases. Of the more common extrapulmonary malignancies, the kidney is the most likely primary site to give lung metastases followed (in descending order) by skin, breast, thyroid, pancreas, prostate, stomach, uterus and colon. Some of these cancer types are, of course, much less common than others and when pulmonary metastatic disease is encountered it is most likely, as a generalization, to be from the breast, followed by colon, pancreas, stomach, skin, kidney, ovary, prostate and uterus. Tumours of the genitourinary and gastrointestinal systems each account for around 28% of lung metastases, the breasts 17% and sarcomas 11%. Metastatic disease may adopt a variety of different patterns in the lung, which may be evident radiologically as well as at gross and microscopic examination of the lungs. Most frequently, metastases present as multiple bilateral nodules. Nodules are usually of variable size with most less than 3 cm diameter, they predominate in the mid to lower zones peripherally and grow rapidly. Metastases may be very large ('cannonball') or may be small and of uniform size (miliary). The following are considered atypical characteristics of metastases: solitary mass, cavitation, central situation, endobronchial location, hilar / mediastinal nodal enlargement, apical tumour with Pancoast or Horner's syndrome, all features suggestive of primary tumour. Atypical cases collectively account for up to 9% of all cases, yet are not infrequently encountered clinically since pulmonary metastatic disease is so common. Pulmonary metastatic disease may also present as diffuse lung consolidation mimicking pneumonia, as interstitial disease due to diffuse lymphatic permeation (lymphangitis carcinomatosa) or direct interstitial infiltration, or with cor pulmonale or other signs of pulmonary vascular compromise due to extensive tumour embolism. Metastases may selectively involve the pleura, mimicking malignant mesothelioma. The Clinical Problem In practice, difficulty most often arises in making a distinction between primary and metastatic disease when there is a solitary pulmonary nodule (SPN) or mass, especially when other atypical features are also present. Primary lung carcinoma may, of course, give rise to lung metastases and lung cancer may co-exist with other non-metastatic pulmonary nodules so that a multiplicity of lesions does not guarantee an extrapulmonary origin for lung malignancy. Solitary metastases account for 3 – 9% of all cases of pulmonary metastatic disease. About 30 – 40% of these are colorectal in origin. The remaining more common sources, in descending order of frequency, are breast, kidney, bladder, testicular teratoma, nasopharynx and sarcomas. Filderman et al [4] considered the problem of the patient with an SPN and a history of extrathoracic malignancy. If the history was of cancer of the breast, stomach, prostate or head & neck (or there was also a past history of lung cancer) then the SPN was most likely primary lung cancer. If the patient had a melanoma or sarcoma, the SPN was more likely a metastasis while a history of renal, colorectal or testicular tumour left an even chance of either diagnosis. Surgical resection of solitary lung metastases is now commonplace and this subject has been recently reviewed [5]. Endobronchial metastases may occur as a result of haematogenous or endolymphatic spread to the airway wall or by invasion into the bronchial lumen by an extrabronchial lesion. Thus definitions of 'endobronchial' vary, as does the reported incidence of this pattern of disease (-1 – 70%). True endobronchial metastases probably account for between 2 – 5% of pulmonary metastases. Breast, colorectal and renal cell carcinoma, plus melanoma and sarcomas account for most cases. Cavitation, present in 9% of primary cancers which are often, though not exclusively, squamous cell is also seen in 4% of metastases, especially those from the head & neck, genitourinary tract, oesophagus and rectum. Once a metastasis is established in the lung it may grow within the lung in a pattern similar to primary disease. Although most metastases tend to have better-defined margins than primary tumours this is not universal, and metastases, especially from pancreas, colon and stomach, may grow in a bronchioloalveolar carcinoma (BAC)-like manner. Metastatic pulmonary tumour may coexist with hilar and/or mediastinal lymph node disease. In patients who present with lymphangitis carcinomatosa, around a quarter will have primary tumour, in Western populations the breast accounts for most cases while in east Asia a gastric primary seems most likely. Solving the clinical problem In this era of multidisciplinary team working the pathologist faced with the diagnostic problem may well be aware of any previous history of malignancy and any clinical or radiological features which point to metastatic disease. Indeed many patients with a clear history of extrathoracic malignancy and typical multiple intrapulmonary nodules on radiology may not be investigated provided the context is correct. As alluded to above, the main problems arise in the following scenarios: Small biopsies or cytology samples arrive with the pathologist without any history indicating a likely / suspected extrathoracic origin for the tumour or the number of nodules present. A solitary pulmonary nodule or mass is biopsied or resected and no history of previous extrathoracic malignancy is given, or metastatic disease is not suspected. Thus, in many instances the pathologist is at the mercy of the requesting physician / surgeon, relying on an adequate history being provided. Armed with a full history including radiological findings, the pathologist can estimate the probability of a SPN being primary or metastatic, using the data provided above, before any microscopic examination. Microscopic examination does not guarantee that even clinically suspected metastatic disease will be recognized. There are many different histological types of primary lung cancer and within each type, many patterns or variants occur. Carcinomas from most extrapulmonary sites have histological mimics in primary lung cancer. As ever, the problems of histological distinction between primary and secondary disease are exacerbated when only small biopsy samples are available. In these circumstances it is often only possible, on the H&E stained sections, to go as far as diagnosing small cell or non-small cell carcinoma, not otherwise specified. Clinicopathological correlation is needed to determine whether disease is primary or metastatic. Sometimes particularly specific features are present, even in small samples, to raise suspicion of metastatic rather than primary disease. These would include melanin pigment in melanoma, nuclear features of papillary thyroid carcinoma, colloid production by follicular thyroid carcinoma and so on for a range of increasingly rare tumours. Unfortunately the majority of lung metastases are from adenocarcinomas which, on H&E examination have features quite acceptable within the broad range seen in primary lung adenocarcinoma, especially when only small samples are available for examination. As a general rule, if a tumour appears microscopically unusual in some way it is worthwhile considering the possibility of metastatic disease and seeking further clinical information and histological evidence to help confirm the nature of the tumour (see below). Features such as clear cells or cribriform adenocarcinoma with 'dirty' necrosis may suggest renal cell or colorectal adenocarcinomas, but may also be seen in primary lung cancer. In resection specimens of peripheral primary adenocarcinoma the mixture of patterns of tumour (solid, acinar, papillary and bronchioloalveolar tumour, often with central scarring) may help identify the lesion. If such a tumour is accompanied by other foci of synchronous adenocarcinoma, including localized non-mucinous BAC and, not infrequently in this situation, multiple AAH, then a primary diagnosis is even more secure. AAH has been found with metastatic disease but usually lesions are singular or few in number. While adenocarcinomatous metastases can show a BAC-pattern, most tend to show homogeneous histology. Squamous cell carcinomas look the same regardless of origin. The presence of squamous carcinoma-in-situ may be a feature in favour of primary disease but CIS and metastatic disease could theoretically co-exist and secondary tumour may invade adjacent airways in a way which mimics in-situ disease. Primary and secondary neuroendocrine tumours also look the same with no reliable morphological features to help make the distinction. In the context of multiple carcinoid tumours, the presence of neuroendocrine cell hyperplasia and tumourlets would support primary disease (DIPNECH). In any case where metastatic disease is suspected, special histological techniques may be used to help confirm the diagnosis. Only in rare circumstances might electron microscopy be useful, but standard special stains for mucin, for example may, if positive, help rule out metastasis from a non-mucin producing primary such as renal cell carcinoma. While genetic profiling has the potential to assist (see below), immunohistochemistry (IHC) is the key technique in this area of differential diagnosis. Most contemporary practice is an evolution of strategies using newer antibodies as they become available. There are markers that may be employed to target certain primary sites if known, but a panel of antibodies may assist in identifying metastatic carcinoma of unknown origin. As with their H&E appearances, squamous cell carcinomas have no IHC profiles specific to particular primary sites. In the context of the carcinomatous SPN which is adenocarcinoma or undifferentiated, the author has used for a number of years a basic, 'first-line' panel of antibodies including TTF-1, CK7, CK20, CEA and PSA or ER and PR. It is also important to take into account any available clinical history and the histological appearances as a guide to the origin of a metastasis and using this information create a more specific, targeted approach to solving the problem. Staining for calcitonin in medullary thyroid carcinoma, thyroglobulin in other thyroid cancers, melanoma-associated proteins, markers to distinguish rarer small round cell tumours from small cell carcinoma, and even lymphoid markers to identify an 'epithelioid' lymphoma may be definitive. It is not possible in this short paper to neither provide a detailed review of the extensive literature on this subject, nor consider every possible extra-pulmonary tumour that might enter a differential diagnosis but some brief comments on the commoner differential diagnoses are given below. Some useful references are provided at the end of this paper. Whichever metastatic site is being considered in the differential diagnosis, the IHC strategy employed must address the possibility that the tumour is primary to the lung. PE10 for surfactant and markers of Clara cells such as DC-LAMP may be useful, but the most frequently used 'lung' marker is TTF-1. Although expressed in thyroid carcinomas and both pulmonary and non-pulmonary neuroendocrine tumours, TTF-1 is expressed in around 75% of lung adenocarcinomas and 50% of large cell and sarcomatoid carcinomas. Metastases from colorectal carcinoma (CRC) are the most frequent lung secondaries encountered. Positive recognition of metastatic CRC is helped by typical histology but this enteric pattern of carcinoma is well described in primary lung adenocarcinoma. Also, not all CRC are typical and, as with other situations, when the H&E histology is atypical so may be the IHC findings. Nonetheless, there is considerable literature developing in this area [6, 7, 8, 9, 10, 11, 12, 13]. The key markers in this particular differential are CK7, CK20, CDX2 and TTF-1. Mucins MUC1, 2 and 5AC have a less well-defined role [14, 15]. Between 72 – 100% of CRC express CDX2 and the majority are CK20+ / CK7-. In around 20% CKs 7&20 are co-expressed, occasional tumours are CK20 - / CK7+. In general, CRC is negative for TTF-1 although there are reports of some positivity with certain antibody clones. MUC2 tends to be expressed in CRC. Most lung adenocarcinomas are CK7+ / CK20- and between 2 – 14% are reported as CDX2+. The difficulties arise when certain unusual patterns of lung adenocarcinoma are encountered which contain goblet cells and on H&E staining mimic CRC to a greater or lesser extent: Mucinous BAC could be confused for metastatic CRC but is negative for CDX2, 17 – 30% are TTF-1+ and while all are CK7+, 60 – 90% also show CK20 positivity. So-called 'colloid' mucinous adenocarcinoma of goblet cell type is more of a problem in that all reported cases have been CDX2+, MUC2+ and 55% CK20+. However, some respiratory phenotype is retained since 82% are CK7+ and 73% TTF-1+. The rare signet ring-cell pattern retains more 'lung' features being CDX2- / MUC2- and positive for TTF-1, CK7 and MUC5AC. Primary lung adenocarcinomas with an enteric pattern including cribriform, gland-in-gland architecture and 'dirty' necrosis do occur. These are indistinguishable on H&E staining from CRC, but again, the lung immunophenotype appears preserved with TTF-1, CK7 and MUC5AC expression while CDX2, CK20 and MUC2 tend not to be found. Surgical resection is now the treatment of choice for pulmonary oligometastatic disease from CRC. In resected cases that show vascular (non-lymphatic) invasion and aerogenous spread with floating cell clusters in alveoli adjacent to the tumour mass, the prognosis was much poorer when compared to cases lacking these features [16]. Renal cell carcinomas frequently show clear cells but so may lung cancer. Both organs may be the site of sarcomatoid carcinoma. Renal cell carcinomas do not generally produce mixed acidic and neutral mucins; most lung adenocarcinomas do. Antibodies to CD10 and so-called renal cell carcinoma (RCC) marker may stain RCC, and alpha-methyacyl-coenzymeA racemase (AMACR) is expressed in some papillary RCCs [17]. These markers are not well expressed in sarcomatoid RCC. Prostate specific antigen (PSA) and prostatic acid phosphatase (PAP) are fairly specific for prostatic carcinoma. Breast carcinomas express ERα and PR. PR lacks specificity and is of limited value in the metastatic setting. ER is more specific but lacks sensitivity [18]. Some lung cancers express ERβ. Gross cystic disease fluid protein 15 (GCDFP15) is very specific (99%) for breast carcinoma but lacks sensitivity (50%). Urothelial bladder carcinoma may be a difficult problem to differentiate but expression of thrombomodulin, uroplakin III and co-expression of CK7 and 20 may help, but like in lung cancer, CEA can be strongly expressed. Testicular teratomas may be easy to recognize if differentiated heterologous or primitive elements are present but undifferentiated embryonal carcinoma may pose problems. Embryonal carcinomas express CD30 and Placental Alkaline Phosphatase. Choriocarcinoma is also easy to recognize but βHCG expression helps. It is interesting that, in reviewing the above markers with which most surgical pathologists would be familiar, expression profiling on a large number of primary adenocarcinomas suggests that panels of antibody markers using many of the above can be effective (88% accuracy) in determining the origin of the 'unknown primary' [19]. A recent study using real-time quantitative PCR suggests a 92-gene assay can give similar accuracy (87%). Interestingly, in the study, the accuracy in identifying lung adenocarcinoma was poor (63%) but was high in lung squamous cell carcinoma (90%) [20]. Genotype profiling has been suggested as an effective method of determining whether a tumour is a new primary or a metastasis from known earlier malignancy by comparing the genotype of both lesions [21, 22]. References Dail DH. Metastases to and from the lung. In: Pulmonary Pathology, 2nd edition. Eds Dail DH, Hammar SP. Springer Verlag, New York, 1994. Pg 1581-1616 Corrin B. Secondary tumours of the lungs. In: Pathology of the Lungs. Ed. Corrin B. Churchill Livingstone, London, 2000. Pg 573-584 Dail D et al. Metastases to the Lung. In Travis WD, Brambilla E, Muller-Hermelink HK et al (Eds) World Health Organisation Classification of Tumours. Pathology and Genetics of Tumours of the Lung, Pleura, Thymus and Heart. IARC press: Lyon 2004. Filderman AE et al. Pulmonary and pleural manifestations of extrathoracic malignancies. Clin Chest Med 1989;10:747-807 Kondo H et al. Surgical treatment of metastatic malignancies. Pulmonary metastasis: indications and outcomes. Int J Clin Oncol 2005;10:81-85 Barbareschi et al. CDX2 homebox gene expression is a reliable marker of colorectal adenocarcinoma metastases to the lungs. Am J Surg Pathol 2003;27:141-149 Saad RS et al. Usefulness of CDX2 in separating mucinous bronchioloalveolar adenocarcinoma of the lung from metastatic mucinous colorectal adenocarcinoma. Am J Clin Pathol 2004;122:421-427 Rossi G et al. Primary mucinous (so-called colloid) carcinomas of the lung: a clinicopathological and immunohistochemical study with special reference to CDX2 homeobox gene and MUC2 expression. Am J Surg Pathol 2004;28:442-452 Tot T. Identifying colorectal metastases in liver biopsies: the novel CDX2 antibody is less specific than the cytokeratin 20+/7- phenotype. Med Sci Monit 2004;10:139-143 Yatabe Y et al. CK20 expression, CDX2 expression, K-ras mutation and goblet cell morphology in a subset of lung adenocarcinomas. J Pathol 2004;203:645-652 Mazziotta et al. CDX2 immunostaining as a gastrointestinal marker: expression in lung carcinomas is a potential pitfall. Appl Immunohistochem Mol Morphol 2005;13:55-60 De Lott LB et al. CDX2 is a useful marker of intestinal-type differentiation: a tissue microarray-based study of 629 tumours from various sites. Arch Pathol Lab Med 2005;129:1100-1105 Yousem SA. Pulmonary intestinal-type adenocarcinoma does not show enteric differentiation by immunohistochemical study. Mod Pathol 2005;18:816-821 Lau SK et al. Differential expression of MUC1, MUC2 and MUC5AC in carcinomas of various sites: an immunohistochemical study. Am J Clin Pathol 2004;122:61-69. Tsuta K et al. Comparison of the immunophenotypes of signet-ring cell carcinoma, solid adenocarcinoma with mucin production, and mucinous bronchioloalveolar carcinoma of the lung characterised by the presence of cytoplasmic mucin. J Pathol 2006;209:78-87. Shiona et al. Histopathological prognostic factors in resected colorectal lung metastases. Ann Thorac Surg 2005;79:278-283. Skinnider BF, Amin MB. An immunohistochemical approach to the differential diagnosis of renal tumours. Semin Diagn Pathol 2005;22:51-68 Nash JW et al. The utility of estrogen receptor and progesterone receptor immunohistochemistry in the distinction of breast carcinoma from other tumours of the liver. Arch Pathol Lab Med 2003;127:1591-1595 Dennis JL et al. Markers of adenocarcinoma characteristic of the site of origin: development of a diagnostic algorithm. Clin Cancer Res 2005;11:3766-3772 Ma X-J et al. Molecular classification of human cancers using a 92-gene real-time quantitative polymerase chain reaction assay. Arch Pathol Lab Med 2006;130:465-473 Tang M et al. Microsatellite analysis of synchronous and metachronous tumours: a tool for double primary tumour and metastasis assessment. Diagn Mol Pathol 2003;12:151-159 Rolston R et al. Distinguishing de novo second cancer formation from tumour recurrence. Mutational fingerprinting by microdissection genotyping. J Mol Diagn 2001;3:129-132