Problems in the Assessment of Tumor Size: An Elusive Grail in Current Practice

Tumor Registrar Errors 

In many hospitals, a tumor registrar will immortalize tumor size by recording and tabulating this important feature from the available pathology report. In the absence of a given tumor size, the size of the biopsy is not infrequently recorded in default as the actual tumor size. When this occurs, it results in erroneous tumor registry reports as well as inaccurate quality assurance activities—both of which are intended to provide appropriate feedback to clinicians to improve patient care.

Eyeball Estimate 

If the tumor sizes as recorded in the hospital tumor registry are tabulated, one will find that there is a clustering around 10, 15, 20, and 25 mm. Only 20% of the numbers 1 to 100 should be evenly divided by 5, yet in our own hospital data from 1993 (Table 2), 60% of all recorded tumor sizes could be evenly divided by 5. Clearly, many of these sizes were estimated pathologically. Although a disposable plastic metric ruler may be the least expensive instrument employed in pathology practice, it is useless if no attempt is made to apply the ruler to the specimen.

Table 2. Frequency of Estimated Tumor Size⁎

	Year	1992	1993	1994	1995
	N	63	38	55	70
	Percent divisible by 5	60	54	51	28

⁎ 20% of tumor sizes from 1 to 100 mm can be evenly divided by the number 5, eg, 5, 10, 15, 20 mm. Percentages greater than 20% likely reflect “rounding off” estimates of tumor size not measurements.

A gross measurement can be unreliable, even when a ruler is employed (Table 3), since this measurement may include adjacent areas of noninvasive carcinoma, proliferative breast disease or even reaction from a prior biopsy. This error results in increasing the prognostic tumor size, and may result in patients being over-treated.

Table 3. Potential Errors in Gross Tumor Size Determination

	• Inclusion of in situ and/or benign disease and post-biopsy reactions in measurement
	• Tumor sectioned on its short axis minimizing tumor size
	• Underestimated tumor size: non-palpable or grossly invisible disease

Sectioning Errors 

A second area in which gross tumor size can be misrepresented results from the common practice of pathologist sectioning (ie, “bread loafing”) a resection at right angles to its long axis (Fig. 1). Particularly when this is done without knowledge of the size(s) established by preoperative imaging, or when the pathologist fails to calculate the size based on a sequential series of sections containing the tumor, a much smaller prognostic size can be recorded. This result also impacts treatment decision making.

View Large Image Download to PowerPoint

Figure 1. Schematic illustrating lumpectomy (white) and tumor (black). Greatest size is parallel to the longitudinal axis but pathology practice is to section at right angles to the long axis of the specimen (bread-loafing) which may result in substantially smaller tumor sizes being recorded.

Addition Errors 

A third error in determining gross or microscopic tumor size results from summing the sizes of the carcinoma in multiple separate morcellated fragments of a resection, or in the multiple separate fragments of a core biopsy procedure. The American Joint Commission on Cancer Staging Manual Sixth Edition8 prohibits this type of calculation, which results in much larger prognostic sizes (Fig. 2). Using the same logic would result in a small sliced crabapple expanding into a cantaloupe.

View Large Image Download to PowerPoint

Figure 2. Schematic illustrating the common error of adding the maximum size of a cancer in the primary excision, with the maximum size of the residual cancer in the re-excision (or mastectomy).

For the majority of image-detected invasive breast cancers, these errors in over or underestimating tumor size are made possible either because of ignorance or indifference to preoperative imaging studies. Such correlation is required to confirm an adequate excision and tumor size, particularly for small image-detected lesions.9, 10 It is difficult to cite a 40-mm tumor size when the pathologist is aware of the fact that the preoperative mammogram and ultrasound only established a 16-mm carcinoma.

There are situations when a definitive resection for a previously biopsied small T1 carcinoma reveals either no residual disease, or only microscopic tumor. In these circumstances, the maximum extent of the invasive component in the prior core biopsy material should be correlated with the preoperative imaging to establish the most accurate estimate of tumor size.

Two actual examples of miscalculated tumor size will illustrate the point: A 46-year-old developed an interval spiculated mammographic density in the left breast, and underwent a needle localized excisional biopsy. The surgeon morcellated the specimen into six fragments, each of which was separately inked by the pathologist. The pathologist estimated a large T2 or T3 size, based on adding the sizes or extents of invasive duct carcinoma in the largest of the two fragments (some 4.5 cm), or all four of the involved fragments (7.3 cm). Oblivious to the implications of a morcellated specimen, he noted multiple margins were transected (Table 4, Table 5; and Fig. 3). The patient was presented to a prospective tumor board, which concluded that induction chemotherapy would be most advantageous before mastectomy. The mastectomy specimen revealed only a small post excisional scar, and no evidence of tumor involution. Review of the preoperative imaging at that point documented a T1c-sized carcinoma in both mammographic and ultrasound studies (Figure 4, Figure 5). At the very least, the patient underwent an unnecessary mastectomy, and probably a more rigorous adjuvant regimen than she would otherwise have been offered.

Table 4. Case One

	Breast, left-radiographically directed biopsy:
	A.Lesion: Infiltrating ductal carcinoma
	B.Grade: Poorly differentiated, Bloom-Richardson: grade II/III
	C.Size: Between 4.5 and 7.3 cm, with positive surgical margins - see comment

Table 5. Case One

	Comment: All four of the biopsy pieces contain extensive infiltrating duct carcinoma with a very minor component of in situ disease. Differentiation varies, with tubule formation in some areas and poorly differentiated tumor elsewhere. Focal Indian filing mimicking infiltrating lobular carcinoma is also present. There is widespread infiltration of regional adipose tissue and margins of each of the four fragments are involved. The minimum size is estimated at 4.5 cm and the maximum size if 7.3 cm or greater, as multiple margins are positive for infiltrating carcinoma.

View Large Image Download to PowerPoint

Figure 3. Morcellated excision biopsy illustrating transected carcinoma at an inked margin showing electrocautery artifact. This cannot be assumed to represent a true margin because of the fragmentation of the biopsy. This margin may represent an internal surface.

View Large Image Download to PowerPoint

Figure 4. Pretreatment mammographic image demonstrates a tumor size consistent with T1c/T2.

View Large Image Download to PowerPoint

Figure 5. Pretreatment ultrasound image demonstrates a T1c size of 16 mm. Neither the mammographic or ultrasound images are consistent with the estimated pathologic tumor size of 4.5 to 7.3 cm. Correlation with imaging is mandatory to accurately establish tumor size for image-detected breast cancer.

A 52-year-old developed an interval left breast spiculated density, measuring 7 mm in maximum size. An ultrasound examination confirmed a solid hypoechoic mass with acoustic shadowing, measuring 7.3 mm in size. An ultrasound-guided core biopsy confirmed an intermediate grade invasive duct carcinoma, Scarff-Bloom-Richardson 6, and a subsequent needle localization and sentinel node procedure revealed an “11 mm” tumor size and node negative status, as recorded in the synoptic report (Table 6, Table 7). This measurement is based on an innovative method of estimating tumor size, which includes biopsy site changes. Chemotherapy was recommended on the basis of the synoptic report recording a size greater than T1b (11 mm). Correlation with preoperative imaging permitted the patient, a physician’s wife, to forgo adjuvant chemotherapy. The maximum size of the invasive carcinoma in the initial core biopsy fragments was 5.0 mm, and in the needle localization only a 0.6-mm microscopic residuum was detected (Figs. 6 and 7).

Table 6. Case Two

	FINAL MICROSCOPIC DIAGNOSIS:
	1. Focal residual infiltrating ductal carcinoma, right mastectomy and sentinel and other lymph node biopsy
	a. Size: 1.1 cm
	b. Grade: Moderately differentiated, Bloom-Richardson grade II/III
	c. Surgical Margins: Free
	d. Vascular/Lymphatic Invasion: Not recognized
	e. Lymph Nodes: No evidence of metastasis to two axillary/sentinel lymph nodes (0/2)
	f. Prognostic factors by immunoperoxidase stain (primarily on prior core biopsy):
	i. Estrogen receptor protein: POSITIVE
	ii. Progesterone receptor protein: NEGATIVE
	iii. Her-2/neu membrane protein by Dako HerceptTest: NEGATIVE (0+)
	g. AJCC TMN: pT1c N0(i-) Mx

Table 7. Case Two

	Comment: There is only focal residual infiltrating ductal carcinoma as basically a strand of cells near the prior biopsy site. I’d estimate the overall area of involvement including the prior biopsy change is just over a centimeter in dimension.

View Large Image Download to PowerPoint

Figure 6. A single 0.6-mm strand of tumor cells was identified in the needle localization. The pathologist calculated an 11-mm tumor size by including the reactive and reparative changes secondary to the ultrasound-guided core biopsy.

Microscopic tumor size is more accurate than gross size measurements, since for a substantial number of invasive carcinomas adjacent areas of DCIS (proliferative breast disease and/or reaction from a prior biopsy procedure) can be excluded with certainty.

Radiology–Pathology Correlation 

Two consensus conferences on image-detected breast cancer9, 10 have concluded that the best estimate of tumor size requires correlation of the sizes seen in preoperative imaging and subsequent pathology. Such correlation is especially relevant for clinically occult lesions, which form the largest group of newly detected breast cancers at the current time. Such correlation requires that the pathologist at least have access to the record of the preoperative imaging, and the specimen radiogram, and this correlation should be documented in the report.

Such correlation is an interdisciplinary process which is more consistently and successfully achieved when it is part of an ongoing review in which all the relevant disciplines actively collaborate. Such correlation is difficult to realize when the various specialists—radiologist, surgeon, pathologist—are segregated in separate noncommunicating facilities.

Tissue Processing 

Another problem area in pathology practice impacting size determination relates to methods of tissue processing. Ideally, resections for small image-detected breast cancers, whether invasive or in situ, and for a prior diagnosis of atypical hyperplasia should be processed in their entirety in conjunction with imaging. This presupposes an intact, oriented specimen which is selectively inked to preserve orientation. The specimen should be x-rayed before inking and fixation. The fixed specimen should be sectioned into uniform sequential segments to preserve orientation. A second specimen x-ray at this point can assist in mapping out the distribution of an invasive carcinoma or of duct carcinoma in situ.11 For DCIS and atypical ductal hyperplasia (ADH) in particular, the entire resection should be processed in a manner which permits some form of three-dimensional reconstruction. Only by mapping out the distribution of duct carcinoma in situ in the sequentially sectioned specimens can the size of the DCIS be calculated, particularly when it is only marginally associated with microcalcification. Additionally, microinvasion or larger areas of invasion may not be discovered by sampling a resection in a manner designed for large palpable carcinomas. This sequential and complete tissue processing methodology was the basis for early studies establishing the likelihood of invasion related to the size or extent of DCIS, and the feasibility of breast conservation for that disease,12, 13 and the later development of the Van Nuys Prognostic Index.14, 15, 16 In contrast, it is the absence of this methodology which prevents a number of prospective randomized trials of radiation therapy for DCIS16, 17, 18, 19, 20 to define prognostically distinct subsets based on size, margin status, and the exclusion of invasive disease.

Conclusions 

Maximizing the accuracy of tumor size determination for the majority of T1 N0 image-detected invasive breast carcinomas, will provide the best estimate of distant disease-free survival (DDFS). A more accurate measurement may preclude unnecessary adjuvant interventions and their associated morbidities and costs for many women, reserving such treatment for those who will benefit. The methods reviewed to achieve an accurate tumor size impose negligible burdens of time, labor and materiel, but do require a consistent pathologic protocol and some planning.

For duct carcinoma in situ patients who undergo segmental resection, there are major benefits in preparing a completely and sequentially processed specimen. The size or extent of DCIS can be calculated by mapping even when it is not associated with microcalcifications. All margins based on an approximate 2.5- to 3.0-mm segment thickness can be evaluated and measured, and most importantly the sequential technique minimizes the chance of missing focal invasion. This technique permits subset analyses which if utilized properly have the potential of sparing upwards of 60% of mammographically detected DCIS all adjuvant therapies, specifically irradiation and Tamoximen. Arguments that this approach is too expensive ignore the actual costs of the alternative stratagem of providing every patient radiation and Tamoxifen: $16,000 per patient versus a pathology workup generating on average 40 blocks—$330 per patient. These small investments have the potential to significantly improve treatment choices and reduce possible therapeutic morbidities and over-treatments for many patients.