Safety margins are considered to be the required distance from a given tumor that will theoretically result in complete excision of the tumor; safety margins are used in surgical planning.1 These margins usually are derived from studies conducted to evaluate outcome related to histologic margins or outcome, compared with the macroscopic safety margins used during surgery. The concept of a safety margin has many confounding factors; however, understanding the relationship of these planned surgical margins to histologically determined margins is necessary to better interpret outcomes related to these measures.1–3
Histopathologic margins are often the standard used to measure success of surgical resection and have been associated with the prognosis for a patient.3,4 The working definition of a histopathologic margin or histopathologic safety margin is the distance from the cut edge of the surgical sample to the closest identifiable tumor cell.5,6 Although histopathologic margins are often considered a defined and objective measure, there are many sources of error and subjectivity that can add uncertainty to interpretation of the results. For cutaneous tumors, separation of deeper anatomic layers from the skin, type of tumor, surgical method, postoperative handling, method of histologic sectioning, and various patient factors can all influence measurements of histopathologic margins.1,2,6–8 Also, the phenomenon of shrinkage (ie, change in sample size between presurgical measurements and posthistologic processing) has been described,8–11 and this phenomenon further complicates interpretation of histopathologic margins in relation to planned surgical margins.
Most of the studies conducted to evaluate skin shrinkage after surgical excision have been performed on samples obtained from humans. Factors reported to influence the amount of skin shrinkage include patient age, location of the tumor, inherent elasticity of the skin in that patient, and presence of scar tissue.10 However, there have been conflicting results that indicate age, sex, and skin lesion type (benign vs malignant) do not affect shrinkage.9 It was once commonly believed that histologic processing, specifically formalin fixation, may affect skin shrinkage. However, studies in human9,10 and veterinary12 medicine have found that formalin fixation has little or no effect on change in sample size.
Shrinkage of samples of clinically normal skin of dogs has been evaluated in 2 studies.3,12 Investigators of the earlier study3 found that the degree of shrinkage is dependent on location and reported that inclusion of a muscle layer but not a fascial plane may decrease the amount of shrinkage. Investigators of the latter study12 evaluated sample size, location, and tension lines; however, in contrast to the earlier study, they did not find that location affected skin shrinkage. Although the earlier study included histologic processing, effects of processing were not assessed independent from effects of formalin fixation. The effect of histologic processing on size of skin samples was not assessed in either of those studies.
The purpose of the study reported here was to examine the relationship between planned surgical and histopathologic margins by evaluating factors (eg, size of skin sample, anatomic location, and histologic processing) that may affect shrinkage. We hypothesized that the degree of shrinkage of histologically processed clinically normal skin samples differs on the basis of sample size and location and that shrinkage occurs following histologic processing, but to a lesser degree, than the shrinkage that occurs between excision and formalin fixation.
Materials and Methods
Sample
Skin samples were obtained from 15 canine cadavers consisting of 13 pit bull-type terriers and 2 mixed-breed dogs (6 sexually intact males, 6 sexually intact females, and 3 neutered males). Dogs of breeds with thin skin (eg, Greyhounds) or excessively loose skin were excluded. Cadavers were restricted to dogs with a body weight between 15 and 35 kg and a body condition score of 4 to 6 on a scale of 1 to 9.13 The skin was required to have a grossly normal appearance with no evidence of neoplasia, scars, current lesions, or previous dermatologic disease. The cadavers were obtained from a local animal shelter, and samples were collected within 2 to 8 hours after dogs were euthanized. All dogs used in this study were euthanized for reasons unrelated to the present study.
Sample collection and processing
Templates were created to serve as a guide for excision of skin samples. Templates consisted of durable water-resistant foam and represented 10- and 30-mm margins from a central marker. For the 10-mm samples, length and width of the ellipse were 40 and 20 mm, respectively. For the 30-mm samples, the length and width of the ellipse were 120 and 60 mm, respectively. These dimensions were chosen because they replicated lateral margins created from a central point (eg, tumor), with the length being 2 times the width. Because the experimentally defined tumor in this study was a pinpoint, the sample size was considered equal to half the diameter of the width. This elliptical shape is commonly used clinically to increase the ease of skin closure.
A 10-mm sample and a 30-mm sample were collected from each of 3 anatomic locations (head, hind limb, and lumbar region) on each cadaver (Figure 1). Sample size (10 or 30 mm) was randomly allocated via coin toss to a side for each cadaver; all 10-mm samples were collected from one side of a cadaver, and all 30-mm samples were collected from the other side of that cadaver. Cadavers were placed in lateral recumbency for collection of skin samples from the hind limbs and lumbar region and in sternal recumbency for collection of skin samples from the head. Hair was clipped from each location prior to collection of samples.
Photographs of the locations for the collection of skin samples from the head (A), hind limb (B), and lumbar region (C) of canine cadavers. In panel A, notice the placement for the 30-mm template on the left side of the head (right side of photgraph) and the 10-mm template on the right side of the head.
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
Photographs of the locations for the collection of skin samples from the head (A), hind limb (B), and lumbar region (C) of canine cadavers. In panel A, notice the placement for the 30-mm template on the left side of the head (right side of photgraph) and the 10-mm template on the right side of the head.
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
Photographs of the locations for the collection of skin samples from the head (A), hind limb (B), and lumbar region (C) of canine cadavers. In panel A, notice the placement for the 30-mm template on the left side of the head (right side of photgraph) and the 10-mm template on the right side of the head.
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
Samples from the head were centered over the temporal region, with the long axis of the ellipse in a cranial-to-caudal direction. For samples from the lumbar region, the center was at the midpoint between the last rib and the crest of the wing of the ilium at a point approximately 4 to 5 cm lateral to the vertebral spinous processes; the long axis of the ellipse was in a cranial-to-caudal direction. For skin samples collected from the hind limbs, the patella and greater trochanter were detected by use of manual palpation, and the sample was centered between these 2 points. Samples were obtained cranial to the long axis of the femur, and the long axis of the ellipse was parallel to the line of tension.14 Direction of the ellipse was chosen on the basis of a consistent fascial layer with minimal muscle attachment, to facilitate bilateral positioning for collection of the samples, or both.
The template was lightly held in place so that the skin was pulled or stretched as little as possible. By use of the template, an ellipse was drawn on the skin with a fine-point marker, and a needle coated in surgical inka was used to mark the center (Figure 2). The template was removed, and the margins were measured for accuracy by use of a soft ruler accurate to 1 mm.
Photographs illustrating steps involved in processing of skin samples obtained from canine cadavers. A—An elliptical area was drawn on the skin with an indelible marker; the black circle in the middle represents a tumor. A partial-thickness skin incision was made; length and width were measured at this stage (prior to excision [P1]). B—The incision was extended deeper into the skin until the fascial layer could be identified; depth of the sample was measured (P1). Ink dots of different colors were placed on the sample for orientation. C—A full-thickness incision was made on one side of the sample, and the sample was undermined beneath the fascial layer. Sutures were placed in the sample, and the sample was then excised. D—Excised skin samples were placed in neutral-buffered 10% formalin, and measurements were obtained after fixation for 24 to 48 hours (P2). E—Skin samples were embedded in paraffin. F—Slides of tissue samples were used for measurement of length and width (after histologic processing [P3]).
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
Photographs illustrating steps involved in processing of skin samples obtained from canine cadavers. A—An elliptical area was drawn on the skin with an indelible marker; the black circle in the middle represents a tumor. A partial-thickness skin incision was made; length and width were measured at this stage (prior to excision [P1]). B—The incision was extended deeper into the skin until the fascial layer could be identified; depth of the sample was measured (P1). Ink dots of different colors were placed on the sample for orientation. C—A full-thickness incision was made on one side of the sample, and the sample was undermined beneath the fascial layer. Sutures were placed in the sample, and the sample was then excised. D—Excised skin samples were placed in neutral-buffered 10% formalin, and measurements were obtained after fixation for 24 to 48 hours (P2). E—Skin samples were embedded in paraffin. F—Slides of tissue samples were used for measurement of length and width (after histologic processing [P3]).
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
Photographs illustrating steps involved in processing of skin samples obtained from canine cadavers. A—An elliptical area was drawn on the skin with an indelible marker; the black circle in the middle represents a tumor. A partial-thickness skin incision was made; length and width were measured at this stage (prior to excision [P1]). B—The incision was extended deeper into the skin until the fascial layer could be identified; depth of the sample was measured (P1). Ink dots of different colors were placed on the sample for orientation. C—A full-thickness incision was made on one side of the sample, and the sample was undermined beneath the fascial layer. Sutures were placed in the sample, and the sample was then excised. D—Excised skin samples were placed in neutral-buffered 10% formalin, and measurements were obtained after fixation for 24 to 48 hours (P2). E—Skin samples were embedded in paraffin. F—Slides of tissue samples were used for measurement of length and width (after histologic processing [P3]).
Citation: American Journal of Veterinary Research 77, 9; 10.2460/ajvr.77.9.1036
A partial-thickness skin incision through the epidermis was made along the marked ellipse (Figure 2). A photograph of the ellipse and ruler was obtained. Photographs were subsequently used to calculate SA. Length (long axis of the ellipse) and width (widest point of the ellipse perpendicular to the long axis) were measured prior to excision of the samples (ie, first processing stage; P1). Initially, 3 measurements were obtained and the mean value was calculated for use; however, this practice was discontinued because of the highly repeatable nature of the length and width measurements. After the length and width were measured, the incision was extended deeper into the skin at the points used to measure the length and width until the fascial layer could be identified. A depth gauge typically used for cortical bone measurements was used to measure the P1 depth (distance from the surface of the skin to the fascial layer). The most cranial, caudal, lateral (or medial), and dorsal (or ventral) extents of the incision were marked with different colors of surgical inka; the ink was allowed to air dry for 10 minutes. A full-thickness skin incision was made on half of the sample, and the remainder of the deep fascial layer was undermined. The skin was loosely sutured to the fascia at 3 of the ink-marked points to prevent movement of fascia prior to completion of the skin incision and sample removal. Once the sample was excised, a fourth suture was placed at the remaining ink-marked point. Samples were placed flat on the bottom of a sample container and immersed in neutral-buffered 10% formalin (approx 1 part tissue to 10 parts formalin) for 24 to 48 hours. Measurements as described for P1 were again obtained 24 to 48 hours after the start of formalin fixation (ie, P2). The same investigator (JKR) performed all measurements at P1 and P2. After the length and width measurements were obtained at P2, photographs were obtained and subsequently used to calculate SA.
Samples were trimmed via radial cuts. The first cut was made on the short axis between the ink-marked points. Samples were then cut on the long axis between the ink-marked points. The trimmed samples were subjected to standard embedding in paraffin. Slides were stained with H&E stain and evaluated by a board-certified veterinary pathologist (KS). Measurements of the long and short axis for each of the trimmed sections for each slide were added to determine the length and width, respectively, at this processing stage (ie, P3). Because of tissue sectioning and creation of artifacts during histologic processing, depth was not measured at P3.
Photographs obtained at P1 and P2 were used to calculate SA. Publically available softwareb was used for SA calculations.
Statistical analysis
Descriptive statistics were calculated for body condition score and body weight. Continuous variables were assessed for normality by analyzing histograms for skewness and kurtosis and by use of the Shapiro-Wilk test. Mean and SD were reported for normally distributed variables, and median and range were reported for nonnormally distributed variables. Percentage change for length, width, depth, and SA between P1 and P2 was calculated by use of the following equation: ([P1 measurement – P2 measurement]/P1 measurement) × 100. Percentage change for length and width between P2 and P3 was calculated by use of the following equation: ([P2 measurement – P3 measurement]/P2 measurement) × 100. Percentage change for length and width between P1 and P3 was calculated by use of the following equation: ([P1 measurement – P3 measurement]/P1 measurement) × 100. All calculations were performed with commercially available software.c
Linear mixed models of commercially available statistical softwared were used to evaluate the effect of time, anatomic location, and sample size on changes in length, width, and SA. Length, width, and SA represented fixed effects, and dog was a random effect. This statistical method was used because of the loss of 1 data point for SA. Significance was set at values of P < 0.05.
Results
Samples
A total of 90 samples were obtained from the 15 cadavers (6 samples/cadaver). Median body condition score was 5 (range, 4 to 6), and mean ± SD body weight was 23.6 ± 3.9 kg. All measurement data were available for all samples, except for 1 missing data point for the SA calculation of 1 dog because of a lost photograph.
Change in absolute length and width of samples
Time, anatomic location, and sample size all influenced length, width, and SA. Mean and SD of the length and width for each anatomic location and sample size were summarized (Tables 1 and 2).
Mean ± SD absolute length at various stages of processing* for 2 sizes of skin samples obtained from 3 anatomic locations of 15 canine cadavers.
| Sample size (mm) | Anatomic location | P1 length (mm) | P2 length (mm) | P3 length (mm) |
|---|---|---|---|---|
| 10 | Hind limb | 40.9 ± 1.4a | 31.1 ± 2.3a | 26.3 ± 2.4a |
| 10 | Lumbar region | 40.0 ± 1.1b | 33.1 ± 2.9b | 26.7 ± 2.9b |
| 10 | Head | 40.9 ± 1.4c | 35.3 ± 2.3c | 27.9 ± 4.7c |
| 30 | Hind limb | 121.7 ± 3.2a | 84.4 ± 4.5a | 75.9 ± 2.2a |
| 30 | Lumbar region | 120.5 ± 3.5b | 96.7 ± 6.5b | 85.3 ± 7.0b |
| 30 | Head | 121.1 ± 2.9c | 102.9 ± 8.0c | 92.1 ± 11.2c |
Within a row, all values differ significantly (P < 0.001).
Measurements were obtained prior to excision (P1), after samples had been fixed in neutral-buffered 10% formalin for 24 to 48 hours (P2), and on slides prepared for histologic evaluation (P3).
a–cWithin a sample size within a column, values with different superscript letters differ significantly (P < 0.001).
Mean ± SD absolute width at various stages of processing* for 2 sizes of skin samples obtained from 3 anatomic locations of 15 canine cadavers.
| Sample size (mm) | Anatomic location | P1 width (mm) | P2 width (mm) | P3 width (mm) |
|---|---|---|---|---|
| 10 | Hind limb | 18.9 ± 0.9a | 17.9 ± 1.6a | 15.5 ± 1.5a |
| 10 | Lumbar region | 20.9 ± 1.1b | 17.7 ± 0.9b | 16.1 ± 1.6b |
| 10 | Head | 20.9 ± 1.0c | 17.1 ± 0.9c | 16.9 ± 1.8c |
| 30 | Hind limb | 59.5 ± 1.7a | 49.3 ± 2.9a | 47.2 ± 4.2a |
| 30 | Lumbar region | 61.3 ± 1.8b | 51.1 ± 2.1b | 47.3 ± 3.5b |
| 30 | Head | 61.3 ± 1.6c | 50.9 ± 3.4c | 47.9 ± 4.1c |
See Table 1 for key.
Absolute length and width decreased significantly (P < 0.001) from P1 through P3 for both sample sizes and all anatomic locations. There were significant (P < 0.001) decreases in length and width between P1 and P2, P2 and P3, and P1 and P3. Generally, the decrease in length and width between P2 and P3 was less than that between P1 and P2 for the 3 anatomic locations; these changes were more marked in 30-mm samples.
Length and width of all samples at the 3 anatomic locations decreased significantly (P < 0.001) over time (P1 through P3) for both sample sizes. Hind limb samples had a significantly (P < 0.001) greater decrease in length, compared with results for samples obtained from the lumbar region and head, between P1 and P2, P2 and P3, and P1 and P3 for both sample sizes. Samples from the lumbar region had an intermediate decrease in length, and samples from the head had the smallest decrease in length; results for each of these anatomic locations differed significantly (P < 0.001) from results for all other locations at all processing stages and both sample sizes. The absolute change in width was similar for both sample sizes for all locations among all 3 processing stages.
Sample length decreased significantly (P < 0.001) for both 10- and 30-mm samples between PI and P2, P2 and P3, and P1 and P3 for all anatomic locations. Decrease in length was significantly (P < 0.001) greater for 30-mm samples than for 10-mm samples at all anatomic locations and processing stages. The decrease in sample width between P2 and P3 was less than that between P1 and P2 for all anatomic locations and both sample sizes.
SA
Mean and SD of the SA for each anatomic location and sample size were summarized (Table 3). The SA decreased by 10.3% to 59.3% between P1 and P2 for all samples. Mean ± SD percentage change for the SA for the 10-mm samples from the head, lumbar region, and hind limb was 33.5 ± 12.6%, 37.5 ± 9.8%, and 40.0 ± 10%, respectively. Mean percentage change for the SA for the 30-mm samples from the head, lumbar region, and hind limb was 34.1 ± 9.9%, 39.4 ± 0.1%, and 48.8 ± 6.3%, respectively.
Mean ± SD value of SA at various stages of processing* for 2 sizes of skin samples obtained from 3 anatomic locations of 15 canine cadavers.
| Sample size (mm) | Anatomic location | P1 SA (mm2) | P2 SA (mm2) |
|---|---|---|---|
| 10 | Hind limb | 492.2 ± 47.6 | 293.8 ± 46.4 |
| 10 | Lumbar region | 545.5 ± 67.4 | 337.2 ± 46.8 |
| 10 | Head | 530.1 ± 54.9 | 349.5 ± 57.0 |
| 30 | Hind limb | 4,145.1 ± 335.9 | 2,107.7 ± 160.3 |
| 30 | Lumbar region | 4,229.4 ± 314.0† | 2,504.5 ± 360.4 |
| 30 | Head | 4,197.3 ± 482.2 | 2,751.4 ± 437.8 |
Represents results for only 14 samples because the photograph for 1 sample was lost.
See Table 1 for remainder of key.
Percentage change in length and width
Percentage change in length and width between P1 and P2, P2 and P3, and P1 and P3 was evaluated for each sample size and anatomic location (Tables 4 and 5). Percentages were calculated because they were believed to be a more clinically relevant and intuitive measure than was the absolute change.
Mean ± SD percentage change in length between various stages of processing* for 2 sizes of skin samples obtained from 3 anatomic locations of 15 canine cadavers.
| Sample size (mm) | Anatomic location | Change between P1 and P2 (%) | Change between P2 and P3 (%) | Change between P1 and P3 (%) |
|---|---|---|---|---|
| 10 | Hind limb | 23.8 ± 4.3 | 15.3 ± 6.8 | 35.6 ± 4.8 |
| 10 | Lumbar region | 17.2 ± 6.8 | 19.5 ± 5.8 | 33.3 ± 7.0 |
| 10 | Head | 13.7 ± 4.8 | 18.7 ± 8.6 | 35.6 ± 4.8 |
| 30 | Hind limb | 30.6 ± 3.9 | 10.1 ± 4.8 | 37.7 ± 2.0 |
| 30 | Lumbar region | 19.7 ± 5.5 | 11.9 ± 3.7 | 29.3 ± 5.3 |
| 30 | Head | 15.1 ± 6.0 | 10.7 ± 5.9 | 24.0 ± 8.4 |
Percentage change was calculated by use of the following equations: change between P1 and P2 = ([PI measurement – P2 measurement]/P1 measurement) × 100, change between P2 and P3 = ([P2 measurement – P3 measurement]/P2 measurement) × 100, and change between PI and P3 = ([PI measurement – P3 measurement]/Pl measurement) × 100.
See Table I for remainder of key.
Mean ± SD percentage change in width between various stages of processing* for 2 sizes of skin samples obtained from 3 anatomic locations of 15 canine cadavers.
| Sample size (mm) | Anatomic location | Change between P1 and P2 (%) | Change between P2 and P3 (%) | Change between P1 and P3 (%) |
|---|---|---|---|---|
| 10 | Hind limb | 9.5 ± 3.9 | 9.4 ± 6.7 | 18.0 ± 6.9 |
| 10 | Lumbar region | 15.1 ± 6.0 | 8.1 ± 7.5 | 22.6 ± 7.9 |
| 10 | Head | 14.3 ± 7.6 | 5.1 ± 7.3 | 18.7 ± 9.4 |
| 30 | Hind limb | 17.1 ± 4.4 | 4.2 ± 8.8 | 20.6 ± 8.4 |
| 30 | Lumbar region | 16.6 ± 3.7 | 7.5 ± 4.9 | 22.8 ± 5.9 |
| 30 | Head | 17.1 ± 5.0 | 5.8 ± 5.2 | 21.8 ± 6.7 |
See Tables 1 and 4 for key.
For all locations and sample sizes, the percentage change between P1 and P3 ranged from 24.0% to 37.7% for length and from 18.0% to 22.8% for width (Tables 4 and 5). Mean percentage change in width between P2 and P3 was less than the mean percentage change in width between P1 and P2. In general, the mean percentage change in length was greater than the mean percentage change in width between P2 and P3, relative to their corresponding changes between P1 and P2. Mean percentage change in length or width of all samples was greater between P1 and P3, compared with the percentage change between P1 and P2.
Hind limb samples had a greater difference in the mean percentage change in length between P1 and P2, compared with results for lumbar region samples, which were greater than results for head samples (Tables 4 and 5). Mean percentage change in width between P1 and P2 was similar for all anatomic locations and sample sizes, except for the 10-mm hind limb sample. Mean percentage change in length between P1 and P3 was greater for hind limb samples than for the other locations.
Mean percentage change for 10-mm samples between P2 and P3 generally was greater than that for the corresponding 30-mm samples (Tables 4 and 5). Mean percentage change for the 30-mm samples between P1 and P2 generally was greater than that for the 10-mm samples. Mean percentage change in length and width between P1 and P3 was similar for both sample sizes, although there was greater variation in length for the 30-mm samples.
Percentage change in depth
Mean ± SD percentage change in depth between P1 and P2 ranged from −30.9 ± 22.7% to 6.0 ± 13.6%. In general, depth increased between P1 and P2.
Discussion
Tumor type and histopathologic margins are 2 of the most important factors considered when making treatment recommendations for cutaneous tumors. Unfortunately, although the diagnosis of tumor type may be straightforward, interpretation of histopathologic margins has many complicating factors. Efforts have been made in human medicine to further define the relationship between histopathologic and surgical margins for various tumor types and tissues, including tongue or labiobuccal tissue,15 rectal cancer,16 and nonmelanoma skin cancer.17 However, this relationship is poorly understood in veterinary medicine, with only a handful of studies3,12,18–20 conducted to correlate planned surgical margins to histopathologic margins. Before attempting to define this relationship for various tumor types in clinical settings, it is important to gain a better understanding of the manner in which surgical handling, histologic processing, and patient factors affect shrinkage in anatomically normal canine skin.
In the study reported here, decreases in absolute length and width following excision and histologic assessment (P1 to P3) and shrinkage in relation to sample size and anatomic location were corroborated for the most part by results of other veterinary studies conducted to evaluate shrinkage. The overall percentage shrinkage from the time of excision planning through histologic processing ranged from 24.0% to 37.7% for length and 18.0% to 22.8% for width. This is similar to previously reported findings of a total surface change of 21.2% to 32% for canine skin samples3 and 21% for human skin samples.11 Sample shrinkage is an important consideration when attempting to plan surgical margins. Although these results cannot be directly translated to clinical cases, if these findings were similar for a given tumor type, a 10-mm histopathologic margin may equate to approximately a 13- to 16-mm planned surgical margin (safety margin).
Recommendations for size of planned surgical margins differ greatly depending on tumor type. Margin resections are recommended for some benign cutaneous tumors, whereas at the other extreme, planned surgical margins of up to 5 cm are recommended for vaccine-associated feline sarcomas.21 Except for a recent study,12 investigators of other studies have not evaluated whether size of the sample has an effect on shrinkage. In the present study, we found that histologic processing, anatomic location, and sample size all had an effect on the amount of change in length, width, and SA among processing stages. To our knowledge, this was the first study of canine samples that revealed sample size affected the amount of skin shrinkage. In comparing the commonly planned margins of 10 and 30 mm, length had a greater absolute decrease for 30-mm samples than for 10-mm samples at all anatomic locations and processing stages. When the percentage change was evaluated, percentage change in length between P1 and P2 for the 30-mm samples was greater than the change for the equivalent 10-mm sample (15.1% vs 13.7%, 30.6% vs 23.8%, and 19.7% vs 17.2% for head, hind limb, and lumbar region, respectively). However, when the percentage change between P1 and P3 was evaluated, percentage changes were relatively similar when comparing 10- and 30-mm samples for the same anatomic location.
Each anatomic location had differing degrees of skin shrinkage. The hind limb sample had the greatest decrease in length and width for all processing stages for both sample sizes, whereas the lumbar region samples had an intermediate decrease, and the head samples had the smallest decrease. Investigators in another study3 also found that samples obtained from a hind limb distal to the stifle joint (as opposed to proximal to the stifle joint in the present study) had more shrinkage, compared with shrinkage for sites on the head and thorax, which suggests that this is a real association. This behavior may be related to the inherent increase in elasticity or mobility of the skin necessary for movement in that hind limb region. However, results for the study reported here are in contrast to those of a recent study12 in which investigators found that differences in margins and location had no effect on shrinkage. In that study,12 investigators specifically aimed to evaluate shrinkage related to tension lines; however, the study design differed from that of the present study because they collected circular samples (2, 4, and 6 cm). Possible reasons proposed for the differences in findings related to location included differences in sample excision technique, sample shape, breed, and live animals versus cadavers. Given that similar animals were used in both the aforementioned study12 and the present study, shape, location, and excision technique were the most probable contributors. Skin in the pelvic limb, particularly in the cranial portion of the thigh, is highly mobile, and even mild tension can change the amount of skin removed. Also, evaluation in that other study was only up to the point at 24 hours of formalin fixation (equivalent to P2 in the present study) and did not include histologic evaluation. Because the present study found that histologic processing contributed substantially to the final amount of shrinkage, the lack of histologic evaluation in that previous study12 may also have accounted for the differences in results related to the use of various sample sizes. In the aforementioned study,12 results of total percentage shrinkage (equivalent to between P1 and P2 for the present study) ranged from 10.25% to 18.25% and were similar to our results for the same processing stage when hind limb samples were excluded (13.7% to 19.7%).
In general, the mean percentage changes in length between P1 and P2 were greater for the 30-mm samples at a given location than for the 10-mm samples; however, this pattern was diminished when comparing changes between P1 and P3 for the 30- and 10-mm samples. Because the 10-mm samples proportionally had a greater mean percentage change between P2 and P3, in comparison to the percentage change for the 30-mm samples, this variation was likely attributable to effects of histologic processing. Histologic processing had a major effect on shrinkage. The mean ± SD percentage change between P2 and P3 ranged from 5.1 ± 7.3% to 19.5 ± 5.8% for the 10-mm samples and 4.2 ± 8.8% to 11.9 ± 3.7% for the 30-mm samples. Cutting of samples for histologic examination typically results in the loss of 1 to 2 mm of tissue. It follows that this loss would have a greater effect on a small sample than on a large sample. Another possible factor that may have contributed to the difference between the results was interobserver measurement error.
Histologic processing generally consists of many steps, including fixation in formalin, embedding in paraffin, trimming of tissue, tissue staining, and microscopic evaluation of the tissue.4 Skin shrinkage during formalin fixation has been evaluated in both human and veterinary medicine, and most reports3,9,10,12 indicate that shrinkage as a result of histologic processing is minimal. Because this change reportedly is minimal, it was not directly evaluated in the present study, but it may have contributed to some of the differences between P1 and P3. In addition to the tissue that is lost during cutting, there may be some changes to the conformation of the tissue during histologic preparation of the tissue prior to cutting. These losses are difficult to quantify and will likely differ according to each sample being processed.
As part of the methods used in the present study, locations that were measured were marked with surgical ink in an attempt to decrease the error associated with obtaining measurements at different locations. It was evident during examination of samples after formalin fixation that considerable conformational changes occurred in the samples, which made it challenging to obtain accurate measurements despite precise marking of the measurement locations. Some of these changes included rolling of the sample edges; movement, separation, and rotation between the subcutaneous, fascial, and dermal layers; and general distortion of the sample as a result of contraction. Specific protocols were implemented in the present study in an attempt to limit these changes associated with tissue handling and formalin fixation. Only half of each sample was dissected at a time to ensure the layers remained appropriately aligned. A loose, simple-interrupted suture was placed at the measurement locations before removal of a sample from a cadaver to limit shifting of dermal layers. Samples were placed flat on the bottom of formalin containers, and formalin was gently added to ensure the samples remained as flat as possible. Anecdotally, we found that these practices greatly increased the ability to more consistently measure differences between processing stages. Without use of these practices, it would have been difficult or impossible to obtain consistent measurements. Difficulties encountered in this study and the need to adhere to specific practices highlighted the importance of meticulous sample preparation to ensure that accurate margin interpretation by a veterinary pathologist is possible. Practices regarding perioperative and postoperative handling of tissue samples can impact the interpretation of margins, and further studies are warranted to evaluate these practices. Currently, these practices are not standardized. For example, in some studies18,19 conducted to evaluate mast cell tumors of dogs, the specimens were pinned to cardboard in their approximate original shape prior to formalin fixation. The manner in which changes in postexcisional handling of tissues (eg, pinning) affect shrinkage and margin interpretation is unknown.
In an attempt to further quantify overall shrinkage, SA was compared between P1 and P2. The SA decreased between 10.3% and 59.3% for all samples between P1 and P2; however, the mean ± SD percentage change between samples ranged from 34.1 ± 9.9% to 48.8 ± 6.3%. The major source of this variability likely was related to the method of SA evaluation. The SA was calculated from 2-D images; however, the samples curved with the lines of the cadaver and were not in a flat plane in P1 SA images. Images were centered over the samples to account for this as much as possible in P1 images; however, distortion caused by normal body curvature, compared with flat skin, likely caused considerable variability in SA calculations.
In general, samples appeared to increase in thickness (ie, depth) between P1 and P2 for the 4 measurement points. However, there was a wide range in the mean ± SD percentage change among the measurement points (range, −30.9 ± 22.7% to 6.0 ± 13.6%). Most of the mean values were negative, and a negative number represented an increase in thickness. To the authors’ knowledge, this was the first study in veterinary medicine in which investigators attempted to measure depth at P1. A previous study3 revealed an increase in thickness between stages equivalent to P2 and P3. Because of the distortion (ie, movement or separation between the fascial layers) that occurs during histologic processing, depth was not measured at P3 in the study reported here because it was thought likely to be inaccurate.
Limitations of the present study were related to the nature of basic research and use of cadavers. It was possible that postmortem changes within canine skin could have affected the innate properties of skin and therefore skin shrinkage. Skin samples in this study were obtained within 2 to 8 hours after dogs were euthanized to mitigate postmortem effects. Results of this study were consistent with those of another study3 that involved the use of live dogs, which suggested that the use of cadaveric tissue had a small or no effect on outcomes. Also, use of cadavers allowed collection of a greater number of samples, which decreased inherent statistical error. Another limitation of this study was that it was performed with anatomically normal canine skin and did not account for effects that a tumor may have on skin shrinkage. A recent retrospective study20 conducted to evaluate histologic shrinkage specifically for mast cell tumors found that the mean shrinkage was 35% to 42% and that there was a greater percentage of shrinkage in appendicular than in truncal samples for certain directions. These percentages are similar to but overall greater than those for the present study. This difference may have been a reflection of the effect of the tumor, use of live versus cadaveric tissue, or limitations inherent within a retrospective study design such as variations in record keeping, individual measurement techniques, and tumor morphology.
On the basis of results for the present study, the primary hypothesis was accepted that anatomic location and sample size affected the degree of shrinkage when comparing surgical margins to histopathologic margins. Also, the secondary hypothesis that shrinkage following histologic processing (P2 to P3) was less than between excision and formalin fixation (P1 to P2) was also accepted. Given that investigators of previous studies9,10,12 have found that formalin fixation has minimal effects on shrinkage, this suggests that most of the skin shrinkage likely occurred soon after excision. The next step would be to further define factors that affect margin evaluation, such as evaluating shrinkage in relation to margins for various cutaneous tumor types to determine the role of the tumor in shrinkage. Additional factors to be evaluated that may affect shrinkage include other potential tumor locations and various methods of periexcisional and postexcisional handling of samples.
Acknowledgments
This manuscript represents a portion of a thesis submitted by Dr. Reagan to the University of Illinois Department of Clinical Sciences as partial fulfillment of the requirements for a Master of Science degree.
No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors declare that there were no conflicts of interest.
Presented as a poster at the Veterinary Cancer Society Conference, St Louis, October 2014.
ABBREVIATIONS
| SA | Surface area |
Footnotes
Davidson marking system, Bradley Products Inc, Bloomington, Minn.
ImageJ, version 1.48, 1US National Institutes of Health. Bethesda, Md: Available at: imagej.nih.gov/ij/. Accessed Apr 30, 2014.
Microsoft Excel for Mac 2011, version 14.4.4, Microsoft Corp, Redmond, Wash.
SAS software, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Kopke LF, Bastos JF, Filho JA, et al. Safety margin: an old and relative concept. An Bras Dermatol 2005; 80: 279–286.
2. Becker GD. The many faces of surgical margins. Am J Clin Oncol 2007; 30: 556–559.
3. Reimer SB, Séguin B, DeCock HE, et al. Evaluation of the effect of routine histologic processing on the size of skin samples obtained from dog. Am J Vet Res 2005; 66: 500–505.
4. Kamstock DA, Ehrhart EJ, Getzy DM, et al. Recommended guidelines for submission, trimming, margin evaluation, and reporting of tumor biopsy specimens in veterinary surgical pathology. Vet Pathol 2011; 48: 19–31.
5. Auw-Haedrich C, Frick S, Boehringer D, et al. Histologic safety margin in basal cell carcinoma of the eyelid: correlation with recurrence rate. Ophthalmology 2009; 116: 802–806.
6. Donnelly L, Mullin C, Balko J, et al. Evaluation of histological grade and histologically tumour-free margins as predictors of local recurrence in completely excised canine mast cell tumours. Vet Comp Oncol 2015; 13: 70–76.
7. Giudice C, Stefanello D, Sala M, et al. Feline injection-site sarcoma: recurrence, tumour grading and surgical margin status evaluated using the three-dimensional histological technique. Vet J 2010; 186: 84–88.
8. Hudson-Peacock MJ, Matthews JN, Lawrence CM. Relation between size of skin excision, wound and specimen. J Am Acad Dermatol 1995; 32: 1010–1015.
9. Dauendorffer JN, Bastuji-Garin S, Guero S, et al. Shrinkage of skin excision specimens: formalin fixation is not the culprit. Br J Dermatol 2009; 160: 810–814.
10. Kerns MJ, Carst MA, Olsen TG, et al. Shrinkage of cutaneous specimens: formalin or other factors involved? J Cutan Pathol 2008; 35: 1093–1096.
11. Gregory N, Mulvaney M, Pattison T, et al. Shrinkage of skin excision specimens and downcoding. Arch Dermatol 2003; 139: 542–543.
12. Upchurch DA, Malenfant RC, Wignall JR, et al. Effects of sample site and size, skin tension lines, surgeon and formalin fixation on shrinkage of skin samples excised from canine cadavers. Am J Vet Res 2014; 75: 1004–1009.
13. Laflamme DP. Development and validation of a body condition score system for dogs. Canine Pract 1997; 22(7):10–15.
14. Irwin DH. Tension lines in the skin of the dog. J Small Anim Pract 1966; 7: 593–598.
15. Johnson RE, Sigman JD, Funk GF, et al. Quantification of surgical margin shrinkage in the oral cavity. Head Neck 1997; 19: 281–286.
16. Eid I, El-Muhtaseb MS, Mukherjee R, et al. Histological processing variability in the determination of lateral resection margins in rectal cancer. J Clin Pathol 2007; 60: 593–595.
17. Lane JE, Kent DE. Surgical margins in the treatment of nonmelanoma skin cancer and Mohs micrographic surgery. Curr Surg 2005; 62: 518–526.
18. Fulcher RP, Ludwig LL, Bergman PJ, et al. Evaluation of a two-centimeter lateral surgical margin for excision of grade I and grade II cutaneous mast cell tumors in dogs. J Am Vet Med Assoc 2006; 228: 210–215.
19. Simpson AM, Ludwig LL, Newman SJ, et al. Evaluation of surgical margins required for complete excision of cutaneous mast cell tumors in dogs. J Am Vet Med Assoc 2004; 224: 236–240.
20. Risselada M, Mathews KG, Griffith E. Surgically planned versus histologically measured lateral tumor margins for resection of cutaneous and subcutaneous mast cell tumors in dogs: 46 cases (2010–2013). J Am Vet Med Assoc 2015; 247: 184–189.
21. Phelps HA, Kuntz CA, Milner RJ, et al. Radical excision with five-centimeter margins for treatment of feline injection site sarcomas: 91 cases (1998–2002). J Am Vet Med Assoc 2011; 239: 97–106.
