Pathology innovations challenge conventional DCIS assumptions

September 1, 2007

Large-screen photographs of cotton bolls, snakeskin, cacti, and trees saddled with the overgrowth of new limbs often star in provocative lectures about early-stage breast cancer by Dr. Laszlo Tabar, a pioneer in mammography education.

Large-screen photographs of cotton bolls, snakeskin, cacti, and trees saddled with the overgrowth of new limbs often star in provocative lectures about early-stage breast cancer by Dr. Laszlo Tabar, a pioneer in mammography education.

At talks across the U.S., Tabar demonstrates how state-of-the-art mammography, in combination with modern pathology techniques, is exposing the true nature of the many subtypes of breast cancer. A new understanding is emerging of the increasingly detected ductal carcinoma in situ, which now accounts for more than one-fifth of all breast cancers in the U.S.

"The combination [of mammography and advanced pathology techniques] will revolutionize breast cancer diagnosis," said Tabar, a professor of radiology at the University of Uppsala School of Medicine in Sweden.

Informed by regular use of modern pathology techniques, Tabar envisions the underlying tissue structures when he looks at a mammogram and makes a decision. He educates his audiences by comparing the mammographic findings and matching pathology to common sights in the natural world. Fragmented casting and "snakeskinlike" calcifications, appearing either alone or with a mass on the mammogram, are particularly menacing.

Conventional wisdom holds that DCIS consists of malignant cells that have not invaded other tissue, but Tabar has identified subtypes that he thinks are actually invasive and merely mimic DCIS.

Tabar is a strong advocate of subgross thick-section 3D histology techniques in combination with large-format pathology. The latter involves analysis of much bigger chunks than usual of biopsy tissue on large slides. These techniques enable better visualization of normal breast patterns, which vary enormously between patients, helping to reduce unnecessary callbacks for benign processes, according to Tabar. They also make it possible to correlate mammography, 3D breast MRI, and 3D ultrasound images with pathology findings.

"Accurate measurement of the extent of DCIS has proven difficult with conventional pathologic techniques. MRI and 3D ultrasound create a renewed need for correlative pathology, accomplished with large-format breast pathology techniques," said Dr. Lee Tucker, head of pathology and director of the breast care program at Carilion Breast Center in Roanoke, VA, a U.S. test-bed facility for Tabar's theories.

Tabar considers conventional breast pathology in the era of modern breast imaging techniques outdated. Imaging techniques can demonstrate the entire organ on the film or monitor, while "comparative histology" shows only a 4 to 5-micron-thick, 1.5 x 1.5-cm piece of tissue.

"The type of histologic examination used all over the U.S. is an old-fashioned 200-year-old technique," he said.

There is tremendous resistance to adopting large-format pathology, however. It's viewed as highly impractical, expensive, and akin to throwing away the specialty's standard alphabet. The bigger slides don't even fit on conventional microscopes. Despite the obstacles, a few breast centers that have made the transition say the difficulties are exaggerated and the benefits make it all worthwhile.

"Large-format breast pathology is about much more than the use of larger slides. It eliminates the information gap between breast imagers and pathologists by incorporating imaging data into the pathologist's decision making, while preserving regional anatomy and the complex relationship between DCIS and invasive cancer," Tucker said.

DCIS ON THE RISE

Fifty years ago, when the lumps found were generally palpable, the general term breast cancer worked, according to Tabar. Now, radiologists are increasingly finding nonpalpable lesions 1 to 14 mm in size, including DCIS, and more sophisticated diagnostic techniques are required to make comparisons with imaging and correctly determine suitable patient management.

Breast cancer is now thought to be a progressive disease, and most invasive forms start as in situ cancers. Of the 62,000 in situ cancers predicted in the U.S. for 2007, DCIS accounts for about 82%, according to the American Cancer Society. Another 11% are lobular carcinoma in situ, and other types account for the rest.

Given that about 178,000 invasive cancers are also expected in 2007, DCIS now accounts for about 21% of the invasive and in situ cancers put together, whereas its numbers did not register prior to the 1980s (see graph).

DCIS is sometimes depicted by the consumer press and certain clinician groups as a homogeneous, mostly harmless entity. When the American College of Physicians changed its guidelines for screening women in their forties this year, it noted the dangers of overdiagnosis of DCIS, which "seldom results in death." Many breast imagers and pathologists consider DCIS much like invasive cancer, however, in terms of makeup and heterogeneity.

"There is a tendency to generalize about DCIS. It is a heterogeneous disease with many subtypes described by one old-fashioned term. It's like giving the same name to 10 of your children. It's so simple-minded to say you picked up an innocent something when you don't know what potential it had if you had allowed it to grow. It's hard to tell which cute little kid is going to become a criminal," Tabar said.

Breast pathologist Dr. David Page and colleagues at Vanderbilt University agree (Cancer 2005;103:2481-2484).

"A major barrier to achieving a greater level of understanding among clinicians and patients regarding diagnosis and therapy of DCIS is the erroneous idea that DCIS is one disease," they said.

Patient information websites by reputable sources inform women that DCIS is confined to the duct, while lobular carcinoma in situ (LCIS) is confined to the lobules. But 3D histology techniques have shown that both types actually come from the terminal ductal lobular unit (TDLU), as noted in a consensus report from the National Institutes of Health Breast Cancer Think Tank (Oncogene 2000;19;968-988) (see diagram). Cancer can also be localized in the lobules or ducts.

"Most authorities now agree that both types of tumors arise in the terminal ductal lobular unit. That is, they arise in the terminal ductules or alveoli rather than in the collecting ducts as implied by the term ductal. While the designations are useful, they give a misleading impression about the cellular origins of the tumor," the consensus report said.

Tabar thinks it is important to make this distinction in order to match mammography with pathology. TDLUs and ducts have a very different mammographic appearance.

"DCIS is a misnomer to begin with. The vast majority of what we call DCIS is actually in situ within the lobule, not ductal," Tabar said.

Tabar's research indicates that about 75% of lesions currently classified as DCIS have calcifications, including several types shown in figures: powdery (resembling cotton bolls), crushed stone, and casting type (fragmented casting and dotted casting).

Casting-type calcifications are the most deadly, according to a study of small invasive cancers (1 to 14 mm) by Tabar and others, including Dr. Robert Smith, director of cancer screening at the ACS. The researchers reported that findings associated with casting-type calcifications on a mammogram resulted in the worst prognosis for the patient: 52% survival over 20 years compared with 86% to 100% for other mammographic features (Cancer 2004;101[8]:1745-1759).

Tabar proposes that some casting-type calcifications on a mammogram are benign, but most signify that a process called neoductgenesis is taking place, meaning the cancer is forming new ducts or ductlike structures. When this occurs, the finding is usually diagnosed as DCIS at pathology, even though the process of duct formation is invasive, he said.

The ductlike structures do not at all resemble the TDLUs where most DCIS cases arise, Tucker said. In the pathology community, neoductgenesis is accepted for some diseases, such as prostate cancer, but not in breast oncology.

Dr. Michael Lagios, medical director of the Breast Cancer Consultation Service in the San Francisco Bay Area, questions some of Tabar's conclusions. There is not enough evidence to show that high-grade DCIS is invasive, he said.

"Whether neoductgenesis occurs is debatable," he said.

New ducts sometimes do grow-during puberty, for example-but they are clearly not invasive, he said. Lagios points out that for patients with extensive high-grade disease that is accurately assessed, survival rates are essentially 100%. Treatment success for these patients undermines the argument that the cancers are actually invasive rather than in situ.

Cases in which DCIS does turn fatal could represent a missed invasive cancer at the time of diagnosis, he said.

SHATTERED MONA LISA

As DCIS is increasingly diagnosed, interest in breast conserving therapy for the condition is growing, but this requires accurate assessment of margins. In a study of almost 600 women with invasive ductal carcinoma or DCIS, Tucker and colleagues noted that large-format pathology increased the extent of DCIS detected to 16.1 mm from 6.1 mm using conventional techniques. Results were set to be published in the Journal of Clinical Oncology in August.

Using philanthropic funding, Orange Coast Memorial Medical Center in California recently adopted large-format techniques.

"With large-format histology, you see disease in one slide rather than chopped up in multiple small pieces. With conventional techniques, it's like admiring the Mona Lisa in nine pieces. Every small one is good but never the same as looking at the entire painting," said Dr. Julio Ibarra, medical director of breast pathology at the Orange Coast Memorial Medical Center.

Lagios argues, however, that conventional techniques can provide the same information as large-format pathology. It will just take a more thorough analysis than is currently being performed.

"With DCIS, I can't see it. I can't feel it. It is invisible, and it may not calcify. The only way to examine it is to examine all the tissue," he said.

Lagios showed Diagnostic Imaging a DCIS case in which 20 specimen samples were microscopically examined, but a small 6-mm area of invasion was evident in only one sample. It's like the Vasilopita ritual of the Eastern Orthodox Church, he said. A lucky coin is baked inside a loaf of bread, and on New Year's Day, family members gather and slice the loaf thinly and distribute the pieces one by one.

"Finding invasive cancer is like cutting that bread on New Year's morning. If you only cut three or four slices, the likelihood of getting the coin would be nil. To find invasive cancer, you need to cut the way you cut the lucky bread, in thin slices. And you need to examine each slice carefully," Lagios said.

An outdated system offering paltry payment means most pathologists look at only a few slices microscopically rather than the whole specimen.

"Lack of adequate compensation is a big disincentive for pathologists," Lagios said.

Despite the importance of accurate pathologic assessment, quality is highly variable, he said.

"The good thing about imaging is that there are federal quality standards. There are lots of guidelines for pathologists but no mandated standards. That is the problem with breast pathology in the U.S. today," he said.

By the end of 2007 or early 2008, the American College of Pathology will come out with a new DCIS protocol that calls for microscopic examination of the entire specimen. Institutions accredited by the American College of Surgeons Commission on Cancer will need to comply, according to the ACP.

A multidisciplinary consensus report on state-of-the-art image-detected breast cancer advises microscopic evaluation of the entire specimen and mandatory correlation of imaging and pathology studies (J Am Coll Surg 2005;201[4]:586-597). In some hospitals, pathologists do not see the actual images and radiologists do not see the actual pathology. They just see the reports. This is a big limitation for both specialists, Lagios said. Ideally, both should see each others' work, and cases should be discussed at a weekly tumor board meeting.

OTHER CONTROVERSIES

Radiologists may find themselves in the hot seat for detecting DCIS, but most controversy associated with DCIS detection reflects perceptions about overtreatment following diagnosis, according to Smith. An editorial related to a prominent study in The New England Journal of Medicine this year found that computer-aided detection software has been finding mainly DCIS, which is more "indolent" and has less of an effect on mortality rates than other forms of breast cancer.

To date, most of the mortality benefit from mammographic screening has come from detection of invasive cancers, rather than DCIS. As Lagios points out, however, earlier detection could enable a less invasive treatment strategy, a benefit not measured in terms of mortality.

Others argue that since invasive cancers start as in situ cancers, increased detection should cut invasive cancer incidence rates in the future. Any decline in future incidence due to the detection and treatment of DCIS would be spread out over many years, making it difficult to measure, Smith said.

"There are divided opinions about whether detecting and treating DCIS will result in a significant reduction in the eventual incidence of invasive disease. It stands to reason that there will be some effect on future rates, but it is not clear how much," he said.

Smith and Tabar are coauthors of a study that showed most DCIS is probably progressive. At the prevalence screen, 37% of DCIS cases were estimated to be nonprogressive, and at incidence screen, only 4% were estimated to be nonprogressive (Eur J Cancer 2003;39[12]:1746-1754).

"There is an element of overdiagnosis of DCIS in breast cancer screening, but the phenomenon is small in both relative and absolute terms," Smith wrote in an e-mail to Diagnostic Imaging.

A study of low-grade DCIS that was biopsied and left alone found that half (11 of 28) turned into invasive cancer, though it took three decades for this change to occur. Some consumer organizations question why patients with a DCIS diagnosis are given treatment

for full-blown disease. Most women with DCIS, for example, undergo routine radiation after surgical excision. The study of small invasive cancers published by Tabar and others in Cancer concludes that mammographic features can be used reliably to predict a prognosis and that surgical excision alone is enough for many women with early breast cancers.

Meanwhile, surgeons have developed the University of Southern California/Van Nuys Prognostic Index, a formula that can be used to predict outcomes for women with DCIS. It could help spare many women from unnecessary radiation (Oncologist 2003;3[2]:94-103).

In their Cancer article about the progression of DCIS, Page and others agree.

"Evidence now overwhelmingly demonstrates that DCIS is a spectrum of disease ranging from extensive, high-grade lesions, most likely requiring mastectomy for eradication, to small low-grade lesions, which can be cured effectively by excision alone," they said. "It is also clear that there is a middle ground in which excision plus radiation therapy is valid."

Emily Hayes is feature editor of Diagnostic Imaging