From transitional cell to ruothelial carcinoma
William Check, PhD
When the surgical pathologist looks through
the microscope, his or her attention is focused on the specimen under the lens. But always in the back of the pathologist’s mind are two human presences—the clinician and the patient. In the same way the newly published World Health Organization classification system for interpreting and reporting bladder cancer biopsies focuses on the basics of anatomic pathology—pattern recognition and nomenclature. But every change from the 1973 classification derives from data about patient outcomes. And every change in turn has implications for clinician decisionmaking and patient welfare, and carries responsibilities for communication between pathologists and urologists.
"That is exactly why we addressed the problems with the old system," says Victor E. Reuter, MD, chief of surgical pathology at Memorial Sloan-Kettering Cancer Center. "We felt that our classification should be more in tune with the clinical scenarios. Why call something carcinoma grade one if it almost never is associated with invasion and the risk of disease recurrence rates and progression are both very low? And when it does progress, it does so only after manifesting higher-grade disease."
"This system is not just an exercise for pathologists," says Jonathan Epstein, MD, professor of pathology, urology, and oncology and director of surgical pathology at Johns Hopkins Hospitals. "It can’t be made out of thin air. Ultimately, this scheme must correlate with prognosis. And studies so far show it does correlate with prognosis."
The new scheme also provides a common language. "If we are to make any progress in treatment of bladder cancer in terms of surgery and neo-adjuvant therapy, or knowing when we can stop following a patient post-therapy, people doing studies all have to speak the same language," Dr. Epstein adds. "If you call something a papilloma in one institution, it should be the same everywhere, not a papilloma one place and a carcinoma someplace else." Adopting the new system will fix the "Babel of multiple classification systems" that existed previously, he predicts.
Mahul Amin, MD, director of surgical pathology at Emory University Hospital and professor of pathology, urology, and oncology at Emory University School of Medicine, agrees that patients need consistent diagnoses from one institution to the next. "When diagnoses differ, the urologist has a difficult time managing the patient. They think we are playing a name game. It is important in making a classification system to focus on prognostic, biological, and therapeutic aspects of the disease."
David Grignon, MD, chair of the Department of Pathology at Wayne State University School of Medicine, believes that the new grading system will help urologists make treatment choices. When he spoke at symposiums to urologic oncologists, he could see frustration with the old system. A urologist on the panel would ask how many doctors would give adjuvant BCG to patients with grade-three tumors. Almost all said they would. And almost none would give it to patients with grade-one tumors, he says. "But when they were asked about grade two tumors, about half said they would give BCG and half said they wouldn’t. So clinicians were saying, we really don’t know what to do with your grade twos." The new system is superior to the previous classification, he says, in that it divides tumors into more clinically relevant prognostic groups.
"Our clinical colleagues to a large extent base their treatment on the grade and stage that we find," says Peter Humphrey, MD, PhD, professor of pathology and immunology and associate director of anatomic pathology at Washington University School of Medicine and Barnes-Jewish Hospital. "So it is really critical that they know the meaning of the terms in the new scheme and understand its rationale. To that end, it’s important that we have ongoing discussions with our clinical colleagues about it."
The first major step toward the new classification was taken in 1997, when
a group of urologic pathologists was convened in Washington, DC, to discuss
revisions to the 1973 WHO classification. One major controversy was the number
of grades of papillary tumors. In the WHO system there were four grades: papilloma
(grade zero), and transitional cell carcinoma grades one through three. "When
we called grade-one carcinoma, we understood it did not kill," Dr. Amin says.
"There were influential pathologists who wanted to call grade-one carcinoma
as papilloma. This led to confusion about whether to call them carcinoma or
Proposed at the 1997 session was the creation of a new category to be called "papillary urothelial neoplasia of low malignant potential," or PUNLMP, with intermediate status between benign and carcinoma. Papillomas were retained, and carcinomas were divided into high and low grade, for a total of four tiers.
In 1998 WHO and the International Society of Urologic Pathologists endorsed
the new classification and it was published, with Drs. Epstein, Amin, Reuter,
F.K. Mostofi, MD, and the consensus group as authors (Am J Surg Pathol.
1998; 22: 1435- 1448). This year the new system swept the charts: The WHO blue
book on Tumors of the Urinary System and Male Genital Organs was published
using this system; the fourth series of the AFIP fascicle adopted it; and the
CAP protocols endorsed it. "So now we are all on the same page," Dr. Amin says.
A major impetus for this change was clinical data, Dr. Grignon says. Studies carried out in Sweden showed that some tumors classed as grade-two carcinoma, which had lesser degrees of anaplasia and were not as aggressive as grade three, still had the potential to invade and metastasize. "Now most of those are put into the ’high-grade’ category," he says. This makes the high-grade group "a bit of a compromise," he acknowledges. "But we do want to identify those patients and not take a less-aggressive treatment approach."
In addition to redefining tumor grading categories, the new system describes
the characteristics of the categories much more specifically. And it goes into
much more detail regarding definition of invasion for staging. So it offers
improved guidance about both major pathologic influences on prognosis. One further
semantic change was from "transitional cell carcinoma" to "urothelial carcinoma."
In the older system there were only sketchy descriptions of the grading criteria, according to Dr. Epstein. "Literally there were a few words describing those grades," he says. "As a result, pathology of bladder cancer suffered from a total lack of precision on how to make decisions. We were supposed to look at it almost as a gestalt."
In this near-vacuum, grading became subjective. As a result, what started as a three-grade system turned into a five-grade system. "Pathologists would call something grade two to three, or grade one to two," Dr. Grignon says. "Clinicians often didn’t know what we were talking about." Now, as a byproduct of using words rather than numbers, pathologists are forced to use the new scheme as it was intended. "I have yet to see anyone sign out a specimen as ’low grade-high grade,’" Dr. Grignon says. "Those words don’t make sense."
In addition, under the old system many pathologists took the path of least resistance and lumped most tumors into the middle category, grade two, with relatively few grade one or grade three diagnoses. "There was a temptation for the pathologist, if it was an uncertain or difficult case, to put it into the middle ground," says Dr. Humphrey. Unfortunately, grade two was therapeutically ambiguous. Now any tumors with a finite risk of becoming invasive or killing a patient that would have been shunted into grade two under the old classification are pushed into the high-grade carcinoma category, which should contain most clinically significant tumors. "There will always be borderline cases," Dr. Epstein concedes. "Sometimes it is a struggle. But for the most part pathologists can put bladder tumors into high or low grades."
Dr. Grignon estimates that 20 to 30 percent of tumors are being called high grade under the new classification, whereas in the past only about five percent were called grade three. "There is still sometimes a gray zone," he says. "After all, it is a biological system. But to me it is more clear." Dr. Epstein notes that one advantage of having a higher percentage of high-grade cancers is that it allows urologists to follow more closely a larger group of patients at high risk of progression.
In contrast to the old minimalist grading system, the new classification protocol is more like a complex etching by Albrecht Dürer. It defines grade of noninvasive tumors by two architectural features and six cytological features, says Dr. Epstein. (Grading is less valuable once invasion develops, Dr. Grignon notes, since invading tumors are almost all high grade.) Architecturally, for example, the pathologist asks whether tumor cells are lined up perpendicular to the basement membrane, as in normal tissue, or whether they are, as Dr. Grignon puts it, "jumbled up." With regard to the cells themselves, the pathologist looks for the degree of nuclear pleomorphism or nuclear atypia: nuclear overlapping, different degrees of hyperchromasia with H&E stain, irregular nuclear contours, and varying nuclear sizes. "These are all reproducible features that pathologists are well used to," Dr. Epstein says. "Now we can look at bladder tumors using these specific criteria and come up with more reproducible diagnoses."
Dr. Reuter agrees that these are familiar characteristics, but he says they have never been used consistently in this setting. "At first they will make people feel uncomfortable, since this is not the way they were trained," he notes. In the past, the threshold for calling bladder tumors carcinoma was very low. "Virtually every tumor in the bladder that had a papillary architecture was regarded as carcinoma," Dr. Reuter says. With the new classification, pathologists will be asked to go to medium-power magnification to decide whether a specimen is carcinoma and, if so, whether it is low or high grade.
The prognostic difference is major. Low-grade tumors are those that are clearly malignant but of relatively low risk. They are rarely associated with deeply invasive muscle disease and rarely progress to kill the patient. High-grade tumors, on the other hand, are more likely to invade muscle and pose a survival risk.
Like a socially equitable tax cut, the new system offers relief at both the high and low ends. In addition to moving tumors with moderate risk out of grade two and into the high-grade carcinoma category, it also promotes greater accuracy for low-risk lesions. "It was very important to take the lowest grades of tumor, the most benign-appearing lesions, and remove the word ’carcinoma,’" Dr. Reuter says. For this purpose the category of papillary urothelial neoplasia of low malignant potential was created. PUNLMP tumors are never associated with invasion. They can recur, but they typically do so as low-grade tumors and are never or rarely associated with progression. Dr. Epstein calls them "more of a nuisance than a risk to life expectancy."
PUNLMP resolved a problem caused by the diagnosis of papilloma, a mostly benign lesion. Once a papilloma is removed, there is little cause for worry. However, because of its conservative treatment, strict criteria were used for papilloma, so the diagnosis was rarely made. One unfortunate consequence was that everything else was called carcinoma in the old system. "So some patients with very indolent tumors were labeled as having carcinoma," Dr. Epstein notes, "which has psychological, insurance, and treatment implications." Under the new system the restrictive criteria for papilloma are retained, while creation of the intermediate PUNLMP category provides an alternative to labeling everything else as carcinoma. Now these patients are followed, not lost, without being labeled with a malignant diagnosis.
"Some people don’t like the term PUNLMP," Dr. Grignon says. "In some ways I don’t like it either. But that is not the important thing about this system. From the patient’s standpoint it is more important that we are capturing more of these clinically significant lesions into the high-grade category, so that they will be managed more appropriately."
Dr. Humphrey particularly likes the new division into low- and high-grade carcinoma rather than grades one, two, and three. In general, he prefers a binary system, even though it can be difficult to assign some cases. He too focuses on the clinical benefits. "Data are emerging to support putting carcinomas into high and low grade," he says.
As already noted, the new classification was motivated by existing data. "We
borrowed liberally from original work by Mälmstrom’s group," Dr. Reuter acknowledges.
"In their classification of bladder cancer, they made a point of emphasizing
differences in growth pattern characteristics as well as cytology, which they
used to evaluate what they called the level of disorder in the tumor." Going
from benign to PUNLMP to high-grade malignancy reflects an increasing degree
of disorder. Still, morphology is inherently subjective, Dr. Reuter concedes,
and can vary among practitioners. So it was reassuring when studies appeared
from Holmang, a member of Mälmstrom’s original group, showing that using the
modern classification was clinically useful. (See for example: Holmang S, et
al. J Urol. 1999; 162: 702-707.)
A number of studies have been published showing that in general there are biological
differences between papilloma, PUNLMP, and low- and high-grade tumors in terms
of risk of recurrence and progression, Dr. Epstein says. He has reviewed a number
of them (Crit Rev Oncol Hematol. 2003; 47: 83-89). "In general, what
most studies document is that with papilloma, you take it out and the patient
is cured," Dr. Epstein says. Recurrences do occur but typically remain nonmalignant.
In rare cases, several years after a papilloma is taken out a high-grade tumor
arises, so followup is recommended; it is not yet known how long followup is
necessary. Most studies show that PUNLMPs recur but generally don’t progress.
Low-grade carcinomas progress, but not at as high a rate as high-grade carcinomas.
"A minority of studies say it is not so clear," Dr. Epstein says. This could
be due to differences in applying the system or perhaps differences in patient
Dr. Epstein notes that confusion arose initially because some older studies made a one-to-one correlation between categories in the 1973 WHO system and the new system, equating PUNLMP to the old grade one, low-grade carcinoma to grade two, and high-grade carcinoma to grade three. "That is not how it works," he says, cautioning that this is also a potential pitfall in daily practice.
Not surprisingly, all is not perfect with the new grading system. "Particularly in community-based practices, one shortcoming is that it doesn’t translate well to interpretation of urinary cytologic specimens," says Michael Cohen, MD, head of the Department of Pathology at University of Iowa Health Care. There is no "fix" for this problem. "We still interpret urinary cytologic specimens as we normally would," he says. "It is just that some things-such as papillary hyperplasia—can’t be recognized cytologically or don’t translate well. At some point we will probably have to reconcile that."
For staging, the new classification offers extensive guidance as well. For
all tumors, the most important determination is whether there is invasion into
the underlying bladder wall, and if there is invasion, how deep it goes. Beneath
the epithelial layer, or mucosa, of the bladder wall are a connective tissue
layer, the lamina propria (sometimes called the submucosa), and a muscle layer,
the muscularis propria. When tumor has invaded the muscularis propria, the clinician
usually treats aggressively, typically with radical cystectomy. If tumor has
not invaded the muscle wall, greater preservation of the bladder is often possible
through transurethral resection, usually followed by adjuvant BCG. "So the critical
point in therapeutic decisionmaking is whether there is invasion of the muscular
wall," says Dr. Grignon.
Applying this seemingly simple formula can run up against myriad complex issues.
"Many times it is difficult to assess whether a tumor has invaded at all," Dr. Reuter says. "Distinguishing a noninvading tumor from one that has minimally invaded the lamina propria can be a challenge." He cites three possible causes for this ambiguity. Bladder biopsies are almost never oriented; they are "thrown into a cassette," yielding multiple tangential cuts. Bladder tumors not infrequently have an inverted pattern of growth nests that grow down into the lamina propria and can be confused with true stromal invasion. And if the urologist uses a resectoscope that cauterizes tissue to reduce bleeding, it can introduce artifactual changes in tissue, especially with high current.
In the new classification, specific parameters are provided to help with this determination: the growth pattern characteristic of nests; the reaction of stroma around nests; the presence or absence of retractions (an artifact commonly seen in superficially invasive tumor); the presence or absence of what Dr. Reuter calls "paradoxical differentiation" (the fact that microinvasive cells appear to have more cytoplasm than noninvasive tumor components); and loss of the parallel array of blood vessels commonly seen along the basement membrane of urothelium.
Another fundamental challenge to the staging scheme is the occasional absence of muscle from the specimen, in which case the pathologist cannot assess whether there is muscular invasion. "It becomes very important for the pathologist to make clear in the report whether muscle is present," Dr. Grignon says. "Don’t just say ’There is no muscular invasion.’ Say, ’Muscle layer present in material, no invasion of muscle.’ When there is no muscle present, you need proactively to make that comment."
Muscle may be absent from the specimen if a biopsy or small local resection is superficial. Or sometimes with a fairly large tumor the biopsy may capture tumor that is growing exophytically, so it may seem like abundant tissue but contain only tumor and no bladder wall. Cautery artifact can also make it difficult to recognize muscle. In the case of an invasive tumor where muscle is absent because the initial biopsy was not deep enough, the urologist may repeat the biopsy to be sure muscle invasion is not missed, since the recommendation will probably be for local or conservative therapy.
Let’s say the pathologist has now determined that tumor has invaded into lamina propria and that the muscularis propria is present in the biopsy. The next question is whether tumor has invaded muscle. Here arises another possible source of confounding—becoming mixed up between two separate muscle layers. "A muscle layer is also sometimes present within the lamina propria," Dr. Reuter points out. It is called the muscularis mucosae. Clinical attention was first called to its presence in a publication in the 1980s by Jae Ro, MD, et al, of M.D. Anderson. "Recognizing this layer was important because traditionally the presence of invasion into ’muscle’ is what triggers clinicians to want to treat aggressively," Dr. Reuter says. "Once Dr. Ro identified the muscular layer in the lamina propria, it became clear that the pathologist had to decide what muscle layer was being invaded." Most urologists would treat more conservatively if there was only invasion of the muscularis mucosae and not the muscularis propria. The pitfall is that the muscularis mucosae is not always present. "Dr. Ro found it in 93 percent of cystectomies in whole bladder," Dr. Reuter says. "But what really matters is how often you see it in transurethral resections, in the original biopsy. Probably less than 50 percent of the time," he estimates.
Having negotiated that obstacle, the pathologist may begin to feel home safe. But wait—more hurdles lie within the lamina propria. Even if a bladder tumor does not invade the muscularis propria, and thus remains a T1 lesion, it can be important to determine how deeply it has invaded the lamina propria. Says Dr. Grignon, "Some urologists may recommend cystectomy to some patients whose tumors have only invaded the submucosa." They may want to know about pathologic features that predict whether a T1 tumor is at higher risk of recurring with muscular invasion or developing metastases, such as the depth of invasion in the lamina propria and whether there is invasion of blood vessels or lymphatic channels.
"Depth of invasion has been approached a couple of ways," Dr. Grignon says, one of which is to measure it in millimeters. There are no accepted definitions for microinvasion, he notes, although some have suggested 5 mm or less and others have proposed 3 mm or less. "There are limited data to suggest that patients with very little invasion are at much lower risk of metastases than those with deeper invasion within that layer," Dr. Grignon says. "In the future we may see a more formal definition."
A second approach is to use the thin muscle fibers of muscularis mucosae—when it is present—and divide tumors into those that have invaded only superficial to that layer versus those that have invaded into the thin muscle fibers or beyond. "There are 12 to 15 publications on that topic," Dr. Grignon says. "Most show that defining the depth of invasion using that anatomic marker does separate patients into lower and higher risk of dying of bladder cancer." Using the muscularis mucosae presents technical problems, Dr. Grignon concedes. But, he says, "I do know urologists who will ask me whether the tumor has invaded into the muscularis mucosae."
If connective tissue and muscle can complicate staging, why not fat? Invasion into perivesical adipose tissue increases a tumor’s stage. But, like muscle, adipose tissue exists in more than one location in the bladder. "There has been a recent recognition that adipose tissue can be found in all layers of the bladder," Dr. Humphrey says, "including within lamina propria and muscle wall. So if the pathologist sees carcinoma in fat tissue, that doesn’t necessarily mean that the cancer is outside the bladder."
While papillary lesions are the most common type of bladder tumor, another important presentation is the nonpapillary or "flat" lesion, which has a flat mucosa. Says Dr. Amin, "Conceptually, flat disease is considered a precursor to papillary or invasive cancer." In the 1973 classification, flat lesions were categorized into normal; reactive; mild, moderate, and severe dysplasia; and carcinoma in situ, or CIS. In the new classification these six categories have been reduced to four—all flat lesions with atypia that are not reactive are simply called dysplasia, since the levels of dysplasia did not yield clinically useful information and the interobserver reproducibility between the categories was poor. "A diagnosis of dysplasia tells us that the urothelium is abnormal," Dr. Amin says. "It is a preneoplastic stage." CIS, on the other hand, signifies a neoplastic classification, which is prognostically and therapeutically critical. "Dysplasia is a marker of instability," Dr. Amin sums up, "while CIS needs to be treated."
Aside from these main types of bladder cancer, Dr. Humphrey says, several subtle histological variants have been recognized over the past five years. They are uncommon but important for differential diagnosis. In urothelial carcinoma with giant cells, the giant cells are distinctive and should be distinguished from urothelial carcinoma with foreign body, osteoclast-like giant cells and urothelial carcinoma with trophoblastic differentiation. Micropapillary and lymphoepithelioma-like carcinoma have also been described.
A lymphoma-like variant of urothelial carcinoma presents a pitfall, since it can mimic lymphoma. There are both lymphoma-like and plasmacytoid variants. "As long as the pathologist recognizes that these variants exist, appropriate stains can be ordered to confirm an epithelial malignancy," Dr. Humphrey says.
Urothelial carcinoma also grows in variants that can look like benign disease,
such as microcystic variant and nested variant. "This latter is particularly
important to recognize," Dr. Humphrey says. "While rare, it is potentially aggressive
despite its bland cytological appearance." A constellation of findings aids
recognition: high density, nearly confluent growth pattern and a degree of nuclear
atypia, which may be only focal and found only in deeper portions of the proliferations.
Nested variant urothelial carcinoma must be distinguished from von Brunn nests
(Volmer KE, et al. Am J Surg Pathol. 2003; 27: 1243- 1252).
How widely adopted is the new classification system at this time? Certainly all of the pathologists interviewed for this story are using it. And Dr. Grignon thinks it has been broadly accepted: "I would say the majority of expert urologic pathologists are utilizing it worldwide now. Pathologists who present short courses for USCAP, CAP, ASCP—that is the scheme they are largely presenting. And when I interact with urologists, they are getting it in their pathology reports wherever they are."
Dr. Epstein agrees. "I ask at meetings how many are using it. Five years ago maybe 20 percent said they were; now it’s the majority. But it’s not yet unanimous."
Dr. Amin’s impression, "as I give lectures," is that 60 to 70 percent of pathologists use this classification and another 10 percent use both the 1973 and 2004 systems. "Now it is up to us to understand and apply the system consistently and accurately," he says.
Dr. Humphrey sets one more, rather imposing, task for pathologists. With any new system, he says, it’s important that it be tested prospectively in clinical trials. "The only true test of utility of any pathological grading and staging scheme is using it to assign treatment, which is something often not pursued within the pathology community," he says. He points out that one of the most powerful grading schemes, the Gleason score for prostate cancer, arose from clinical trials. The new bladder tumor classification could be embedded as part of a clinical trial, Dr. Humphrey suggests. Pathologists should become more involved with clinical oncology groups in clinical trial design and in reviewing slides from trials, where pathologic typing, grading, and staging of cancer can be critically important in enrolling and stratifying patients in therapeutic trials.
"Here is where we can make a big contribution," he says.
William Check is a medical writer in Wilmette, Ill.