Context.—Several common differential diagnoses are encountered in urologic pathology, frequently causing patient referrals for a second opinion.
Objectives.—To review 3 common differential diagnoses encountered in a large consultation service in genitourinary pathology, including partial atrophy versus prostatic acinar adenocarcinoma, oncocytoma versus chromophobe renal cell carcinoma, and urothelial carcinoma in situ versus normal urothelium and reactive atypia. We will discuss the detailed, morphologically distinctive features and the usefulness of immunohistochemistry.
Data Sources.—Personal experience and review of the current literature.
Conclusions.—Careful morphologic assessment and awareness of diagnostic pitfalls are fundamental in reaching a definitive diagnosis in most cases. Immunohistochemistry is useful but should be used only in conjunction with the morphologic impression.
The following are 3 of the most common differential diagnoses encountered in urological pathology.
PARTIAL ATROPHY VERSUS PROSTATIC ACINAR ADENOCARCINOMA
One of the most common benign mimickers of prostatic acinar adenocarcinoma in prostate biopsies is partial atrophy (PTAT), and this differential diagnosis is a frequent reason for consultations. In one of the earliest reports describing PTAT, which included 51 cases referred for second opinion, more than one-half of the cases were either called cancer or considered to be possible cancer by the referring pathologist.1 Partial atrophy is 1 of the 3 main types of atrophy, which also include simple atrophy and postatrophic hyperplasia (PAH).2 Simple atrophy appears as a well-circumscribed area of glands of normal caliber that are spaced apart in a configuration similar to that of normal epithelium. In comparison, PAH consists of small, crowded acini arranged in a lobular configuration, often surrounding a central, dilated duct. In both simple atrophy and PAH, the glands appear basophilic, a typical feature seen even at scanning magnification. The basophilia results from the lack of cytoplasm, both apically and laterally compared with normal epithelium, so the nuclei appear crowded and a nuclear outline of the glands is seen at low-power magnification. Simple atrophy and PAH do not generally pose a diagnostic problem. In cases where there is an issue, immunohistochemical stains for basal cell markers (p63, high–molecular-weight cytokeratin) show uniform staining of the basal cells in the glands of simple atrophy and PAH, ruling out prostate cancer. In addition, racemase results are generally negative in simple atrophy and are seldom expressed in PAH.3,4
Partial atrophy differs from simple atrophy and from PAH in several aspects: first, although PTAT generally retains its lobular architectural pattern of growth, that is not always the case, and it can show a more-disorganized, diffuse growth pattern. Second, the glands in PTAT do not have the typical atrophic, basophilic appearance. Instead, PTAT appears as a focus of crowded glands with pale, scant cytoplasm. The attenuated cytoplasm is mostly apical, with most of the nuclei in the glands reaching the full height of the cells. The lateral aspect of the cytoplasm is, however, preserved, which causes nuclei that are more spaced and less crowded than the glands of simple atrophy or PAH, hence, the pale, rather than basophilic glandular, appearance at low magnification. Features seen within PTAT that create difficulty in distinguishing it from prostatic acinar adenocarcinoma include the following: (1) crowded and sometimes disorganized patterns of growth, (2) relatively high nuclear to cytoplasmic ratio with slightly enlarged nuclei, (3) straight luminal borders in some glands, (4) the presence of visible but small nucleoli, (5) negative staining of some glands for basal cells markers, and (6) positive staining of some glands for racemase (Figures 1, A through F, and 2, A through F). The diagnosis of prostate adenocarcinoma and its mimickers should not be based on one feature alone (eg, nucleoli or lack of basal cell markers), but rather on a case's constellation of architectural, cytologic, and ancillary features. In PTAT, the disorganized pattern that can sometimes be seen can give the focus a pseudoinfiltrative appearance in which the smaller glands of PTAT seem to be present between the larger, clearly benign glands with papillary infolding. However, that growth pattern is usually focal, and most important, close inspection reveals that the smaller “suspicious” or “atypical” glands are cytologically similar to the larger, benign glands they appear to infiltrate. In addition, the cytoplasm in PTAT is clear or pale in contrast to the typical amphophilic cytoplasm of adenocarcinoma. However, one should not rely on the cytoplasmic quality alone because many adenocarcinomas also have pale cytoplasm. The high nuclear to cytoplasmic ratio seen in PTAT is, as mentioned, from the lack of apical cytoplasm rather than from the marked nuclear enlargement seen in adenocarcinoma. Nucleoli are usually absent to visible in PTAT and are not as prominent as those seen in adenocarcinoma. Mitoses are absent in PTAT, so detection of mitotic figures favors a diagnosis of adenocarcinoma, although they are seldom seen in adenocarcinoma either. The presence of similar cytologic features between the PTAT glands and the adjacent, recognizably benign glands is reassuring because it points to a benign process. In addition, although some of the glands within a focus of PTAT can have straight luminal borders, in all cases, some of the glands within a focus will show the undulating luminal surfaces and papillary infolding typical of benign glands.1 The gradual transition from glands with these more-benign features to those with straighter luminal surfaces (with both types of glands exhibiting identical nuclear and cytoplasmic features) is more consistent with a benign lesion than with adenocarcinoma infiltrating benign glands. Associated simple atrophy or PAH can be seen either on the same core in which PTAT is present or on different cores. Although the presence of simple atrophy or PAH adjacent to a suspicious focus does not rule out adenocarcinoma, it should prompt a closer inspection of the focus to exclude PTAT. As a general rule, the presence of simple atrophy or PAH in cases with a transition between fully developed atrophy and adjacent “atypical glands” are more likely to represent PTAT than they are to correspond to adenocarcinoma. At the immunohistochemical level, PTAT shows patchy staining of basal cells and, in most cases, no staining with racemase. Some glands may appear totally negative for basal cells markers, but that should not be interpreted as support for an adenocarcinoma diagnosis. Instead, those “negative” glands need to be interpreted within the context of the entire “atypical” focus by comparing the morphology of those negative glands with the surrounding glands that show basal cell staining. It is recommended that glands that are negative for staining in this context be considered benign, as long as they share the same benign cytomorphologic features of the associated surrounding glands in the same focus. Negative basal cell marking in some glands is a well-recognized phenomenon in benign mimickers of adenocarcinoma, especially the mimickers PTAT and adenosis. On the other hand, racemase is often overexpressed in the glands of patients with PTAT and should not be regarded as worrisome as long as the lesion shows otherwise typical PTAT features.
Partial atrophy differs from atrophic acinar adenocarcinoma in that atrophic cancers have one or more of the following features: (1) a more-infiltrative appearance, where the cancer glands infiltrate as isolated glands between benign glands; (2) associated nonatrophic cancer; or (3) prominent cytologic atypia beyond what can be seen in benign atrophy.5 Typically, a focus of atrophic cancer should be supported by negative immunohistochemical stain results for basal cell markers. If a small focus of atrophic glands is worrisome but does not fulfill the above criteria, the lesion should be reported as “Atypical glands, suspicious for adenocarcinoma; a definitive diagnosis cannot be rendered because of the presence of atrophic features.” Another instance in which this “atypical” diagnosis might be used is when an entire focus that resembles PTAT is completely negative for basal cell marker staining, and the pathologist is not 100% confident that the lesion is PTAT. That being said, we rarely order basal cell markers on foci that are typical for PTAT on hematoxylin-eosin stain, and we usually rely on the morphologic features alone to make this diagnosis.
ONCOCYTOMA VERSUS CHROMOPHOBE RENAL CELL CARCINOMA
One of the most common reasons for consultation in renal neoplasia remains the differential diagnosis of renal oncocytoma (RO) and the eosinophilic variant of chromophobe renal cell carcinoma (ChRCC). The distinction between those 2 neoplasms is important because ROs are benign and ChRCCs are low-grade malignant neoplasms that sometimes behave aggressively and carry a risk of recurrence and distant metastases. Despite various publications studying the morphologic, immunohistochemical, and molecular differences between those 2 entities, the number of cases with this differential diagnosis encountered in our consultation service is, in fact, increasing with the advent of new minimally invasive forms of therapy in which preoperative needle biopsies of renal tumors are becoming more common. The small tissue sample in those cases makes it more challenging for the pathologist to render a definitive diagnosis in problematic cases.
The overlapping features of the RO and ChRCC neoplasms start at the macroscopic level. Renal oncocytomas are typically well-circumscribed, solid, homogeneous tumors with a brown-mahogany color and a central scar reported in about one-third of cases. Similarly, ChRCCs are well circumscribed, solid, and homogeneous with a tan to brown color and a central scar that is less commonly (about 15% of cases) seen than it is in RO. The variation of the color of ChRCCs reflects its microscopic composition, with tumors containing mostly cells with eosinophilic cytoplasm displaying a brown color and those with predominant light cytoplasm having a beige-tan color. Renal oncocytomas are typically unencapsulated in comparison to ChRCCs, which can contain a capsule visible either at the macroscopic or microscopic levels.6 Both tumors can show areas of hemorrhage, but macroscopic areas of necrosis should not be seen in an RO but are seen in ChRCCs.
Microscopically, the typical appearance of an RO is of nests of cells present in a loose, hypocellular, myxoid stroma (Figure 3, A). The nests can undergo cystic changes, taking on a microcystic appearance and, less commonly, a macrocystic growth pattern. In other instances, compact, back-to-back nests can give the tumor a solid growth pattern. However, solid architecture is usually focal and admixed with more typical nested areas. The cytoplasm of an RO is granular and deeply eosinophilic, reflecting the abundant, packed mitochondria seen at the ultrastructural level (Figure 3, B and C). In some cases, areas of the tumor are composed of oncoblasts, which are identical to the cells of the usual oncocytoma, except that the cytoplasm is not as abundant (Figure 3, D). Focal areas of cytoplasmic clearing are not uncommon but should be focal and restricted to the areas of the central scar, representing degenerative clearing of the cytoplasm because of ischemic changes (Figure 3, E). Renal oncocytoma nuclei are typically single, although binucleation can be seen, and they are centrally located and have regular nuclear membranes and absent to prominent nucleoli (Figure 3, C and F). Perinuclear halos are absent. Permissible “atypical” features in an RO include hemorrhage, minimal microscopic necrosis, rare typical mitosis, focal clear cell changes (in areas of scarring), entrapped tubules at the periphery of the tumor, and focal areas of atypia.7,8 Although foci with marked nuclear atypia in the forms of hyperchromatic and pleomorphic nuclei can be seen in RO, the atypia is usually degenerative and the nuclei have poorly preserved, smudgy chromatin patterns with associated cytoplasmic intranuclear inclusions and a “blown up” appearance (Figure 3, G). Another form of atypia seen in RO, which can be confused with ChRCC, is crenated, small, dark nuclei that also represent degenerative changes, mimicking the wrinkled, clefted nuclei in ChRCC (Figure 4, A). Focal papillations can sometimes be seen in cystic tubular glands, but prominent papillary features are absent. Additional features that can be seen in ROs are infiltration of perinephric fat and microscopic lymphovascular extension (Figure 4, B and C). However, and despite a recent report of 7 ROs in which macroscopic extension into the renal vein or into its segmental branches was present, it is prudent to carefully resample or reevaluate a tumor that otherwise looks like RO but shows lymphovascular invasion or an extension into a major vascular branch because those features, although not incompatible with a diagnosis of RO, are rather uncommon.9
The eosinophilic variant of ChRCC, which constitutes about one-third of all ChRCCs, is exclusively or predominantly composed of eosinophilic cells, and this variant is the major source of confusion with RO. Chromophobe renal cell carcinoma typically has a solid growth pattern with sheets of cells separated by thin, incomplete, fibrous septa (Figure 4, D). Nested, tubular, cystic, and papillary or pseudopapillary architectural patterns are less commonly encountered. Presence of thick-walled hyalinized blood vessels is also a common finding. The neoplastic cells are usually an admixture of clear and eosinophilic cells, the proportion of which differs from case to case. The clear cells have a pale, frothy, finely reticulated cytoplasm and sometimes appear ballooned out and hydropic. In some cases, a spatial distribution is seen with the clear cells growing around the fibrous septa and the eosinophilic cells located more centrally away from the septae. In other cases, however, a specific distribution is absent, and clear and eosinophilic cells are randomly distributed within the tumor. Perinuclear halos are characteristic and reflect the presence of microvesicles around the nucleus at the ultrastructural level (Figure 3, B). Those microvesicles, which do not have affinity to hematoxylin-eosin staining, are thought to be related to defective mitochondriogenesis. They displace cytoplasmic organelles to the periphery of the cytoplasm leading to prominent cell membranes, a feature typical of ChRCC. Although all the above features are useful in the differential diagnosis, the most important distinguishing features of ChRCC remain its nuclear characteristics: frequent binucleation, irregular nuclear membranes imparting a “wrinkled” or “raisinoid” appearance, coarse chromatin, and prominent nucleoli (Figures 3, B, and 4, E). The finding of notched, irregular nuclei with well-preserved chromatin rules out RO. Mitoses are rare in ChRCC, and necrosis is present in about 20% of cases. In problematic cases in which nuclear atypia is not evident at first glance, scanning the slides with a ×20 objective usually reveals the nuclear hyperchromasia and irregular nuclear membranes of ChRCC that could be missed at lower magnification.
If a Fuhrman nuclear grade were to be assigned to ChRCCs, most would qualify as grade 3/4. That, however, does not reflect their biologic behavior, which is much better than clear cell RCC, to which the Fuhrman grading system was originally described. Therefore, it is not recommended that a nuclear grade be assigned to ChRCC, unless it shows nondegenerative, pleomorphic nuclei that justify the assignment of a Fuhrman nuclear grade 4.
In terms of ancillary techniques that could be helpful in this context, Halles colloidal iron is still used with varying degrees of success in some institutions, despite the test being difficult to perform and having results that vary among different laboratories. In classic cases, Halles colloidal iron staining is diffuse and strong in ChRCC and weak in RO (usually staining only the luminal cellular surface). However, we have seen a number of cases sent for consultation that show overlapping and nonclassic results with this stain (ie, focal staining in an otherwise typical ChRCC or diffuse staining in a typical RO). Therefore, we do not find this stain very useful in distinguishing these entities, and we do not perform it de novo in a problematic case. At the immunohistochemical level, there is a significant overlap between these 2 tumors; they are both negative for vimentin, carbonic anhydrase-IX, and racemase and positive for CD117 and cadherin (E-cadherin or kidney-specific cadherin) and show variable staining results with CD10 and epithelial membrane antigen.10 Recent studies suggest that S100A1 is positive in most ROs and negative in ChRCCs, but those findings remain preliminary.11,12 We find the immunostain CK7 to be most useful. Although most ChRCCs show strong and diffuse staining with luminal accentuation, most ROs are either negative or show scattered CK7+ cells; more CK7 staining is noted in areas of scarring. However, heavy emphasis on CK7 alone in distinguishing RO from ChRCC is discouraged because, again, we have seen classic cases of RO that have stained more diffusely with CK7. On a practical basis, a case that shows convincing morphologic features of ChRCC or RO but conflicting CK7 staining results should be carefully reevaluated for its hematoxylin-eosin morphology, with further sampling if needed. Ultimately, the diagnosis of RO or ChRCC is rendered on the hematoxylin-eosin morphologic impression.
A further confounding factor is that some renal neoplasms show mixed morphologies, bridging RO and ChRCC within the same tumor. This observation supports the theory that RO and ChRCC are related tumors, with these “hybrid” tumors representing the link between the 2 extremes of the spectrum.13 There have been conflicting reports about what constitutes a hybrid tumor. In the first large series on hybrid tumors, Tickoo et al13 reported on a group of patients with renal oncocytosis, some of whom developed tumors with hybrid morphologies. In that study,13 ,hybrid tumors were described as tumors with a background of typical RO in which ChRCC-like areas were present. In most hybrid cases, the transition from RO to ChRCC was gradual, and the transition was characterized by nuclear irregularity and hyperchromasia (wrinkled nuclei), perinuclear halos, or finely reticulated clarity of cytoplasm in an otherwise typical RO-like background (ie, areas with nested architecture and myxoid stroma)13 (Figure 4, F). The other context in which hybrid tumors are frequently encountered is in patients with Birt-Hogg-Dubé (BHD) syndrome, an autosomal-dominant genodermatosis, characterized by fibrofolliculomas, lung cysts, and multifocal renal tumors that include hybrid tumors, ChRCC, clear cell RCC, papillary RCC, and RO.14 In a large series of 130 tumors from 30 patients with BHD, one-half of them were hybrid tumors, which made them the most-common renal neoplasms in patients with BHD syndrome.15 In that report, Pavlovich et al15 defined hybrid tumors as an exclusive solid growth pattern with a mixture of different cell types that were intimately intermixed rather than forming juxtaposed areas/nodules with different histologies. The 2 major cell types are RO-like cells, with eosinophilic cells, poorly defined cell borders, and round, homogeneous nuclei, and ChRCC cells, with clear cytoplasm, well-defined cell borders, wrinkled nuclei, and perinuclear halos.15 The same group16 recently presented, in abstract form at the 2011 United States and Canadian Association of Pathology meeting, the largest series of BHD-related, hybrid tumors. In addition to the typical hybrid, heterogeneous appearance of their previously reported hybrid tumors, some hybrid tumors contained well-formed nodules of typical ChRCC, within an otherwise classic RO. Of note, E-cadherin stained hybrid tumors in a patchy fashion, similar to the RO staining pattern, and the percentage of different cells in a single tumor did not have a clinical effect because no patient with a hybrid tumor as their only renal neoplasm developed metastatic disease16 (Figure 4, G and H). By molecular-clustering analysis, BHD-associated hybrid tumors seem to be located between clusters of ROs and ChRCCs, and by electron microscopy, they show the presence of both microvesicles (similar to ChRCC) and abundant mitochondria (similar to RO).17 Recently, sporadic hybrid tumors (unassociated with either BHD syndrome or renal oncocytosis) have been reported,18 and the description given is of tumors with solid growth patterns of cells that resemble RO cells but have perinuclear halos without wrinkled nuclei. Those tumors had immunohistochemical and ultrastructural features closer to ROs and followed a benign clinical course, albeit with short follow-up.18 In one recent study,19 otherwise typical ROs that had more than 10% of their cells staining for CK7 were reclassified as hybrid tumors. Most would, however, disagree with such an approach because the diagnosis of a hybrid tumor rests on morphologic, not immunohistochemical, grounds. Finally, there have been reports of sporadic ROs with limited foci of ChRCC, which some authors opted to consider as ChRCC (based on the dogma that tumors tend to behave in accordance with their most aggressive component) and others considered as ROs (based on the arguments that those foci are focal and had CK7 and colloidal iron staining pattern similar to RO).8,20
Finally, taking into consideration that ChRCC can have focal areas that resemble ROs, we do not advocate assigning a definitive diagnosis of RO on a needle biopsy. Instead, such cases are signed-out as a renal oncocytic neoplasm, and the following comment is added: “If this biopsy is representative of the entire lesion, it would be consistent with an oncocytoma. However, renal cell carcinoma may uncommonly show focal areas with oncocytic features.” In many of these cases, the lesion sampled on the needle biopsy is ablated by various methods, so the needle biopsy will be the only tissue sampled from the lesion. In cases where the lesion has not been ablated during biopsy, if the needle biopsy morphology is consistent with oncocytoma and the radiologic findings are also consistent, we recommend to the urologist that the tumor is likely an oncocytoma. However, because there is a small possibility that the lesion could be renal cell carcinoma with areas resembling oncocytoma, clinical correlation is required to determine therapy. One of the reasons why it has been difficult to study ChRCC and its distinction from RO is that organ-confined ChRCC has more than a 95% cure rate, so determining the biologic behavior of various oncocytic tumors, such as hybrid tumors or oncocytic tumors on needle biopsies, would require a large study with a large enough number of cases to be able to identify the few cases of metastatic disease.
UROTHELIAL CARCINOMA IN SITU VERSUS NORMAL UROTHELIUM AND REACTIVE ATYPIA
One of the more common differential diagnoses in the bladder leading to a case being sent for consultation is in biopsies to rule out carcinoma in situ (CIS). Our approach has been to, first, analyze the urothelium using the ×10 objective lens. At that magnification, CIS typically shows recognizable, abnormal urothelium with enlarged hyperchromatic nuclei, as opposed to normal urothelium, which contains nuclei that are difficult to visualize, appearing as only small dark “dots.” Careful attention to this low-magnification difference can help prevent overdiagnosing normal urothelium as dysplastic urothelium or CIS. At higher magnification, normal urothelial-cell nuclei do not exceed 3 times the size of normal lymphocytes, whereas, in CIS, nuclei are 5 to 6 times the size of normal lymphocytes. Lymphocytes are usually present in the underlying lamina propria as a reference.
A more difficult distinction is between CIS and reactive urothelial atypia. Before assessing the urothelium, the background presence or absence of associated inflammation should be noted, along with its location and extent. Reactive atypia is typically present when either acute or chronic inflammatory cells infiltrate the overlying urothelium. In contrast, a sparse to moderate lymphocytic infiltrate is almost always present in the lamina propria in biopsies of the normal bladder and does not contribute to significant urothelial nuclear abnormalities. In the presence of inflammation in the urothelium, pathologists should be cautious in diagnosing CIS, although occasional cases of CIS are secondarily associated with inflammation. In addition, a greater severity of intraurothelial inflammation means a greater degree of cytologic abnormality should be allowed. Although lymphocytes involving the urothelium can give rise to reactive nuclear atypia, relatively more-prominent reactive changes are seen when neutrophils are present. Reactive urothelial changes cause uniform nuclear enlargement, nuclear clearing, and central prominent nucleoli (Figure 5, A through C).
Carcinoma in situ is a flat urothelial lesion comprising cytologically malignant cells that involve either the full or partial thickness of urothelium. The urothelium in CIS may be partially denuded, thin, with a normal number of cell layers, or thick. Cellular discohesion is usually more prominent toward the surface causing shedding of CIS cells in the urine. Sometimes the entire urothelium is denuded, necessitating a careful search for residual neoplastic cells. Reactive urothelial atypia is typically not denuded or discohesive, although it may vary in the overall thickness of the urothelium.
In CIS, there is typically a loss of polarity, where the usual arrangement of urothelial cells evenly distributed parallel to each other and perpendicular to the surface is severely altered. However, this feature need not be present because, in some CIS cases, an orderly arrangement is maintained. In others cases, only a few neoplastic cells with enlarged atypical nuclei are scattered throughout the urothelium, for instance, in pagetoid CIS. Within reactive atypia, the polarity is generally maintained. A pagetoid spread of enlarged reactive nuclei would not be seen among normal-sized nuclei because the whole field would be expected to be reactive to the same stimulus.
Nucleomegaly is a consistent feature of CIS, as described above. However, that feature, by itself, is not particularly helpful in distinguishing reactive atypia from CIS because reactive nuclei are also enlarged. The distinction is that, in addition to enlargement, CIS nuclei typically are hyperchromatic and vary in shape and size (Figure 5, D). However, in some cases, the malignant nuclei are uniformly enlarged, yet still hyperchromatic, and diffusely replace the urothelium, which causes a “monotonous” appearance. Another CIS pattern that may be particularly difficult to distinguish from reactive atypia is when prominent nucleoli are present (Figure 5, E). In contrast to reactive atypia, CIS with prominent nucleoli lack associated inflammation, have more hyperchromatic nuclei, and are more pleomorphic. In reactive atypia, the nuclei are associated with inflammation and are uniformly enlarged, vesicular, similar in size and shape one to another, and have a single, central nucleolus. Mitotic figures do not distinguish between the 2 entities because reactive atypia may demonstrate numerous mitotic figures. It is unusual to see atypical mitoses in either condition.
For cases in which, on morphologic grounds, the distinction between CIS and reactive atypia is uncertain, the pathologist should discuss with the urologist whether there are certain clinical coexisting conditions that predispose to reactive changes, such as an indwelling catheter, a colovesicular fistula, or calculi, among others. In these settings, where the morphology is not that of classic CIS, a definitive diagnosis of malignancy should not be rendered. When the distinction between reactive atypia and CIS cannot be made with certainty, the diagnosis “urothelial atypia of unknown significance” can be rendered21 (Figure 5, F and G). These cases typically have intraurothelial inflammation and atypia that is out of proportion to the extent of inflammation, so a neoplastic process cannot be excluded. The message conveyed to the urologist is that the patient should be followed with repeat biopsies after inflammation subsides. This terminology, which is neither a biologic entity nor a disease, should be used sparingly and should not become a “wastebasket category” that includes cases with minimal atypia falling within the range of reactive changes.
There is a spectrum of severity in the architectural and nuclear atypia in CIS, and different patterns have been described:
Large cell CIS with pleomorphism and abundant cytoplasm, which is the easiest form to recognize and shows considerable loss of polarity, discohesion, and significant variation in size and shape.
Large cell CIS without overt pleomorphism or abundant cytoplasm, where the nuclei are uniformly enlarged, imparting a monotonous appearance, which may mimic reactive urothelial atypia and is usually distinguished by the lack of associated intraurothelial inflammation and by paying particular attention to nuclear enlargement, malignant chromatin pattern, and irregular nuclear membranes.
Small cell CIS, which has nuclear features identical to large cell CIS without overt pleomorphism but has cells with scant cytoplasm.
Pagetoid CIS with isolated or small clusters of CIS cells with abundant cytoplasm colonizing a normal urothelium in a pattern similar to mammary Paget disease, which is a pattern usually seen in cases with diffuse multifocal CIS (Figure 6, A and B).
Clinging CIS, which is characterized by a denuded urothelium in which patchy, residual, single carcinoma cells are attached to the basement membrane (Figure 6, C). In cases in which the entire urothelium is denuded, recuts or deeper cuts may identify clinging CIS cells. We recommend ordering deeper sections in cases with an extensively denuded urothelium and with an underlying hypervascular and inflamed lamina propria typical of CIS. If CIS is still not identified, the following comment is added to the case; “denudation may sometimes reflect cases of CIS where the neoplastic cells are shed because of discohesion. Additional tissue or cytology sampling may be helpful if clinically indicated.” 22
In most cases, the distinction between CIS and reactive conditions is straightforward and is based on nuclear characteristics, but in some instances, this differential diagnosis proves difficult. A panel of immunohistochemical stains consisting of CK20, p53, and CD44 has been proposed as useful in making this distinction, and currently, a combined triple stain has been developed and is commercially available. In reactive urothelium, p53 is absent to focal staining, CD44 shows diffuse or patchy staining of all layers, and CK20 stains only the umbrella cell layer. In contrast, the classic staining pattern for CIS is strong and full-thickness staining by p53, absent staining with CD44, and full-thickness staining by CK20.22–24 These findings are mostly based on immunohistochemical correlation in morphologically unequivocal cases of CIS and reactive atypia in studies performed retrospectively. However, how to interpret their results in borderline cases in daily practice is not entirely clear. We have seen classic cases of reactive atypia where CK20 expression has not been restricted to the umbrella cells. Furthermore, p53 staining is difficult to interpret because there is often a spectrum of weak to moderate staining that is nonspecific. Only when the p53 staining is diffuse and intense does it suggest CIS. CD44 has been shown to be the most variable of these 3 antibodies.25 A recent study, available only in an abstract form,26 reported strong and diffuse CK5/6 staining in reactive urothelium and negative or patchy staining (restricted to the basal layer) in CIS, but the utility of this stain remains to be confirmed. Finally, Ki-67 stain should be used with great caution because proliferation indices of reactive and CIS lesions overlap. Therefore, on a practical basis, immunohistochemistry is primarily useful in cases in which the pathologist strongly favors a diagnosis of CIS but has reservations about making a definitive diagnosis. That being said, we do not routinely perform the stains de novo in all borderline cases, but if they were to be done, interpretation of the stains must be made in conjunction with the morphologic impression.
Before rendering a diagnosis of CIS, one should be familiar with the pitfalls in the diagnosis of atypical, flat urothelial lesions (other than inflammatory reactive atypia mentioned above). The following are the major considerations:
In thick and tangentially cut sections, the urothelium may appear crowded and hyperchromatic.
Urothelial cells in the renal pelvis, urethra, and bladder neck have larger nuclei and diminished cytoplasmic clearing and, hence, may be misinterpreted as neoplastic. This issue becomes particularly important in upper tract biopsies, the results of which may trigger nephrectomy if a diagnosis of CIS is rendered. Therefore, caution is advised in interpreting upper tract specimens (this applies to cytologic specimens as well).
Cells in von Brunn nests have slightly larger nuclei than overlying urothelial cells have.
Typical umbrella cells can be displaced so they are not on the surface, mimicking CIS. The key to recognizing umbrella cells is noting their abundant, eosinophilic cytoplasm; their low nuclear to cytoplasmic ratio; their multinucleation; and the absence of hyperchromatic nuclei.
Radiation causes urothelial cell enlargement and sometimes prominent nucleoli. However, maintenance of the nuclear to cytoplasmic ratio; multinucleation; a smudgy, degenerative chromatin pattern; and cytoplasmic vacuolization are characteristic features not seen in CIS (Figure 6, D). Radiation changes can last up to 1 year following treatment, but changes can last longer in the cells of von Brunn nests. Associated hemorrhage, fibrin deposition, atypical stromal fibroblasts or myofibroblasts, and vasculopathy in the lamina propria are associated features that are seen only at later stages, but if present, they could be clues to the reactive nature of the lesion when the history of radiation is not provided.
Bacille Calmette-Guérin therapy induces noncaseating granulomas, lamina propria edema and inflammation, urothelial ulceration, denudation, and regenerative changes with inflammatory-type atypia (uniformly enlarged nuclei with fine chromatin, prominent nucleoli, and variable mitotic activity).
Intravesical chemotherapy (thiotepa [N,N′N″-triethylenethiophosphoramide] and mitomycin C) induces atypical changes, primarily on the surface umbrella cells, which show nuclear enlargement, multinucleation, smudged chromatin, and cytoplasmic vacuolization.
Cyclophosphamide induces urothelial changes similar to those seen in radiation, in addition to hemorrhagic cystitis and possible reactivation of polyoma virus infection.
Polyoma virus infection results in cytologic changes primarily affecting the renal tubules and, rarely, the bladder urothelium. Although the virus causes nucleomegaly, high nuclear to cytoplasmic ratio, and hyperchromasia, the key to reaching the correct diagnosis is noticing the homogeneous, basophilic, smudgy, opaque nuclear inclusion that contrasts with the coarse chromatin pattern in CIS. In addition, nuclear membranes are smooth, and nucleoli are absent. Antibodies directed against the polyomavirus can be used if necessary.
In summary, this article reviews 3 of the most common, challenging differential diagnoses encountered in a large genitourinary pathology consultation service. Careful morphologic assessment and awareness of the diagnostic pitfalls are fundamental to reaching a definitive diagnosis in most cases. Immunohistochemistry is useful but should be used only in conjunction with the morphologic impression.
From the Departments of Pathology and Urology, McGill University Health Center, Montréal, Quebec, Canada (Dr Brimo); and the Departments of Pathology, Urology, and Oncology, Johns Hopkins Hospital Medical Institutions, Baltimore, Maryland (Dr Epstein).
The authors have no relevant financial interest in the products or companies described in this article.