Advertisement

Implant volume estimation in direct-to-implant breast reconstruction after nipple-sparing mastectomy

Open AccessPublished:June 23, 2018DOI:https://doi.org/10.1016/j.jss.2018.05.024

      Abstract

      Background

      Nipple-sparing mastectomy (NSM) is an increasingly popular alternative to more traditional mastectomy approaches. However, estimating the implant volume during direct-to-implant (DTI) reconstruction following NSM is difficult for surgeons with little-to-moderate experience. We aimed to provide a fast, easy to use, and accurate method to aid in the estimation of implant size for DTI reconstruction using the specimen weight and breast volume.

      Methods

      A retrospective analysis was performed using data from 145 NSM patients with specific implant types. Standard two-dimensional digital mammograms were obtained in 118 of the patients. Breast morphological factors (specimen weight, mammographic breast density and volume, and implant size and type) were recorded. Curve-fitting and linear regression models were used to develop formulas predicting the implant volume, and the prediction performance of the obtained formulas was evaluated using the prospective data set.

      Results

      Two formulas to estimate the implant size were obtained, one using the specimen weight and one using the breast volume. The coefficients of correlation (R2) in these formulas were over 0.98 and the root mean squared errors were approximately 13.

      Conclusions

      These implant volume estimate formulas benefit surgeons by providing a preoperative implant volume assessment in DTI reconstruction using the breast volume and an intraoperative assessment using the specimen weight. The implant size estimation formulas obtained in the present study may be applied in a majority of patients.

      Keywords

      Background

      Nipple-sparing mastectomy (NSM)
      • Rusby J.
      • Smith B.
      • Gui G.
      Nipple-sparing mastectomy.
      has gained increased recognition as an alternative to more traditional mastectomy approaches. The acceptance of NSM as a prophylactic procedure, in addition to its therapeutic uses, is increasing.
      • de Alcantara Filho P.
      • Capko D.
      • Barry J.M.
      • et al.
      Nipple-sparing mastectomy for breast cancer and risk-reducing surgery: the Memorial Sloan-Kettering Cancer Center experience.
      In addition, the cosmetic advantages of NSM are significant. Breast volume assessment is one of the most important steps during the preoperative setting in every breast surgery procedure. Appropriately estimating the implant size in NSM is highly dependent on the experience of the surgeons and directly impacts the appearance of the breast after reconstruction. In addition to a common procedure using a temporary sizer during surgery, several replacement solutions, such as using 3D surface imaging and premastectomy breast volume as assessed via mammography or magnetic resonance imaging, have been reported to aid in decision-making.
      • Leibman J.A.
      • Styblo T.M.
      • Bostwick III, J.
      Mammography of the postreconstruction breast.
      • Yip J.M.
      • Mouratova N.
      • Jeffery R.M.
      • et al.
      Accurate assessment of breast volume: a study comparing the volumetric gold standard (direct water displacement measurement of mastectomy specimen) with a 3D laser scanning technique.
      Predicting the breast volume may be useful as an indication for mammoplasty and for calculating the resection weight preoperatively.
      Although breast cancer patients are finding NSM and immediate breast reconstruction surgery to be a highly acceptable option, choosing the appropriate implant volume in direct-to-implant (DTI) reconstruction following NSM is challenging for physicians. In the present study, we aimed to generate an accurate estimate of the implant size after NSM using breast morphological factors (specimen weight, mammographic breast density and volume, and implant size and type). The overall goal of the present study was to provide a fast and accurate method to aid physicians in the estimation of the implant size for NSM patients.

      Methods

      Study population

      This retrospective study was approved by the Institutional Review Board of Changhua Christian Hospital, Taiwan (No. 160110) and was exempt from informed consent. A total of 414 breast cancer patients, aged 30-69 y, who underwent NSM were initially recruited. Patients with bilateral breast reduction or augmentation surgery and patients with incomplete records were excluded. Finally, a total of 182 patients enrolled between January 2009 and December 2015 were included.
      Single-stage DTI and tissue expander/implant reconstructions are the most common implant-based reconstructions after NSM. In consideration of the obtained sample sizes for DTI and tissue expander/implant currently, the present study focused on patients who underwent DTI following NSM. Thus, formula development was performed using data from 145 patients with specific implant types. The patient selection process is shown in Figure 1. Age, weight, height, body mass index (BMI), and associated comorbidities (diabetes mellitus, hypertension, and dyslipidemia) were recorded as patient factors. Specimen weight, mammographic breast density and volume, and implant size and type were recorded as breast morphological factors.
      Figure thumbnail gr1
      Fig. 1Flowchart detailing patient selection. (Color version of figure is available online.)

      Volumetric and density assessment of the breast

      Standard two-dimensional digital mammograms were obtained in 118 of the 145 selected patients using a full-field digital mammography (FFDM) system, including a Senographe Essential/Senographe DS (GE Medical Systems), Mammomat Inspiration (SIEMENS), and Selenia Dimensions (HOLOGIC, Inc). These data covered 62.8% of the study patients and constitute a sufficiently representative data set. All mammograms were obtained from the Department of Medical Imaging at Changhua Christian Hospital in Taiwan.
      Estimating the breast volume from the abnormal side may result in an incorrect implant size. Considering the esthetic value of symmetrical features in breast reconstruction with a consistent standard, the volumetric estimate of the breast was taken from the cranial caudal view of the contralateral side of the lesion (the normal side). For patients with more than one image record, only the record before or on the date of surgery was preserved. Volpara (Volpara version 1.4.2, Matakina Technology, NZ)
      • Highnam R.
      • Brady M.
      • Yaffe M.J.
      • Karssemeijer N.
      • Harvey J.
      Robust breast composition measurement-VolparaTM.
      • Tromans C.
      Comparing Personalized Mean Glandular Dose Estimates Between X-Ray Systems Over Time in Mammography.
      was used to assess the volume and density of the breast and fibroglandular volume on a per-image basis. By analyzing the X-ray dose absorption in the fibroglandular tissue on the raw FFDM data using a specific algorithm, the breast volume was calculated based on the tissue volume.

      Reconstruction after nipple-sparing mastectomy

      This study was conducted in the Comprehensive Breast Cancer Center of Changhua Christian Hospital, Changhua, Taiwan. NSM was followed by immediate implant breast reconstruction. An incision was made on the lateral portion of the upper outer quadrant, 1 cm away from the nipple-areolar complex. At the time of the surgery, a sentinel lymph node biopsy was also performed through this incision. Subcutaneous undermining was then performed along the entire surface of the breast, followed by retro glandular undermining, allowing a complete “subcutaneous mastectomy.” Subnipple tissue was frozen to ensure a cancer-free status. Finally, the mammary gland was replaced by a breast implant under the pectoralis muscle, allowing immediate breast reconstruction. In this way, the breast envelope and the nipple-areolar complex were entirely preserved.
      In all cases, the NSM incision line was located in the upper outer quadrant. The implant was placed under the pectoralis muscle, with or without the serratus muscle elevated inferiorly, depending on the surgeon's preference. Hence, the pectoralis may partially remain subcutaneously along the inferior breast. Currently, acellular dermal matrices are not allowed in Taiwan. The surgeons use a sizer for the implant choice and make decisions based on the visual aspect before the implant placement. The implant volume was recorded for all NSM patients.

      Statistical analysis

      Pearson correlation coefficients between implant size and morphological factors were first examined to determine the most suitable references for implant size choice. Subsequently, curve-fitting and linear regression models were used to develop formulas predicting the implant volume. Least absolute residuals, a robust regression scheme, were applied to minimize the influence of outliers.
      • Bassett Jr., G.
      • Koenker R.
      Asymptotic theory of least absolute error regression.
      • Heiser W.J.
      Correspondence analysis with least absolute residuals.
      IBM SPSS Statistics for Windows, Version 22.0 (IBM Corp, Armonk, USA) was used to generate descriptive statistics. Curve-fitting and linear regression analyses were performed using MATLAB and the Statistics Toolbox Release 2016b (The MathWorks, Inc, Natick, Massachusetts, USA).

      Results

      Patient characteristics

      The patient characteristics and morphological data for the entire data set (n = 182) are shown in Table 1. Over 70% of the patients were aged 40-59 y. The mean BMI of patients aged 40-59 y indicated standard body sizes (40-49 y: 23.08 kg/m2; 50-59 y, 23.7 kg/m2). In contrast, the mean BMI in the elderly patients (aged 60 y or more) indicated mild obesity (60-69 y: 24.36 kg/m2; 70+ y, 24.66 kg/m2). The mean implant volume across all patients was 264.31 cc (standard deviation [SD] = 102.07), and the mean specimen weight was 314.06 g (SD = 178.27).
      Table 1Patient characteristics (n = 182).
      Age (y)BMI (kg/m2)Mean implant size (cc, ±SD)Specimen size (gm, ±SD)n (%)
      <3020.29200.00 (70.71)212.85 (119.01)2 (1.1%)
      30-3922.72290.86 (100.35)326.38 (145.81)29 (15.93%)
      40-4923.08254.06 (91.88)284.96 (134.97)70 (38.46%)
      50-5923.7259.18 (107.71)333.82 (243.37)55 (30.22%)
      60-6924.36270.00 (113.4)328.71 (132.71)24 (13.19%)
      ≥7024.66375.00 (176.78)536.50 (269.41)2 (1.1%)

      Correlations between implant size and breast morphological factors

      The correlation coefficients between the implant volume and breast volume (r = 0.764) and between the implant volume and specimen weight (r = 0.76) were high. In contrast, the fibroglandular tissue volume (r = 0.493) and breast density (r = −0.344) were only modestly correlated with the implant volume. All the P-values in the Pearson correlation tests were less than 1 × 10−3. Given their high correlation with the implant volume, breast volume and specimen weight were selected for formula development.

      Formula development using specimen weight to estimate implant volume

      Although the specimen weight was highly correlated with the implant size, the use of different types of implants (high profile [HP]/moderate plus profile [MPP]) significantly increased the difficulty in generating a prediction model or formula. Among the patients with implant type data, 107 patients (58.1%) used the Mentor (Mentor Worldwide LLC, CA, USA) MPP, 34 patients (18.8%) used the Mentor moderate classic profile (MP), 14 patients (7.6%) used the Allergan (Allergan plc, Dublin, IRL) stylus series, and four patients (2.2%) used the Mentor HP. The type of implant was not recorded in 25 patients (13.6%). Given this distribution of implant types, the prediction model was developed using patients with Mentor MP/MPP/HP implants (n = 145, covering 78.8% of the enrolled patients). The specimen weight and corresponding implant size of these patients were used to generate the prediction model. The linear model had the best prediction performance after the curve-fitting test. The obtained formula was as follows:
      x=0.641y+62.18
      (1)


      where x is the implant size, and y is the specimen weight after NSM (with 95% confidence intervals). The R2 of the formula was 0.9863 (P < 1 × 10−3), and the root mean squared error (RMSE) was 12.52. As seen in Figure 2, this formula is appropriate for most of the patients in the present study.
      Figure thumbnail gr2
      Fig. 2Scatterplot and the linear model regression line for the formula for predicting the implant size using the specimen weight. The specimen weight was measured in 145 selected patients in the retrospective data set. The blue line reflects the formula after the data fitting. (Color version of figure is available online.)

      Formula development using breast volume to estimate implant volume

      Breast volume was also highly correlated with the implant size. Table 2 shows the characteristics of the 118 NSM patients with a Mentor MP/MPP/HP breast implant and concordant mammography. The linear model had the best prediction performance after the curve-fitting test. The obtained formula was as follows:
      x=0.3853y+102.2
      (2)


      where x is the implant size, and y is the breast volume estimated from the cranial caudal view of the mammogram (with 95% confidence intervals). The R2 of this formula was 0.9836 (P < 1 × 10−3), and the RMSE was 13.24, similar to the prediction performance of the formula using the specimen weight. As seen in Figure 3, this formula is appropriate for most of the patients in the present study.
      Table 2Characteristics of the patients with a concordant mammography (n = 118).
      Age (y)Breast volume (cm3, ±SD)Density (VDG, ±SD)Fibroglandular volume (cm3, ±SD)n (%)
      30-39363.38 (190.45)21.5 (7.17)63.2 (25.49)24 (20%)
      40-49412.18 (247.66)18.83 (7.39)63.13 (34.72)50 (41.66%)
      50-59437.72 (249.2)17.33 (8.06)61.38 (34.33)38 (31.66%)
      60-69375.96 (138.57)15.37 (5.05)50.02 (18.36)6 (5%)
      VDG = Volpara density grade.
      Figure thumbnail gr3
      Fig. 3Scatterplot and the linear model regression line for the formula for predicting the implant size using the breast volume. The breast volume was measured in 118 selected patients with a concordant mammography. The breast volume was estimated using Volpara. The blue line reflects the formula after the data fitting. (Color version of figure is available online.)

      Formula development combining breast volume and specimen weight to estimate implant volume

      A two-dimensional polynomial model was created to modeling the breast volume and specimen weight simultaneously, using the estimated breast volume and specimen weight records from 118 NSM patients (the same patients as those used in the development of the formula using the breast volume). The equation was as follows:
      f(x,y)=49.08+0.08332x+0.5551y
      (3)


      where f(x, y) is the function of implant volume, x is the breast volume estimated from the FFDM, and y is the specimen weight (with 95% confidence intervals). The adjusted R2 was 0.976 and the RMSE was 15.55. The fitting surface of this equation is shown in Figure 4.
      Figure thumbnail gr4
      Fig. 4Two-dimensional polynomial data fit for the formula for predicting implant size using the breast volume and specimen weight. The breast volume and specimen weight were measured in 118 selected patients. The breast volume was estimated using Volpara from concordant mammography. The fitting surface reflects the formula after the two-dimensional polynomial data fitting. (Color version of figure is available online.)

      Discussion

      In the present study, two formulas (one based on the specimen weight and one based on breast volume) were obtained for the estimation of implant size, with coefficients of determination (R2) over 0.98 and RMSEs of approximately 13. Thus, a total of 98% of the variance in the implant size for NSM patients can be explained using our models, and the tolerance for the estimated implant size is approximately 13 cc. It is also of interest to determine whether modeling the breast volume and specimen weight simultaneously results in an improved model fit. Hence, a two-dimensional polynomial model was created to answer this question. The larger RMSE and smaller adjusted R2 for the two-variable formula compared to that for the single-variable formulas suggest a slightly lower prediction performance. However, the two-variable formula still obtains a good fit to the data and is provided here as an alternative choice for implant estimation.
      Initially, data from all enrolled NSM patients in the present study were applied to the curve-fitting procedure without any selection. Even after fine, careful adjustments and the exclusion of outliers, the obtained R2 was approximately 0.83 for the specimen weight and was only 0.65 for the FFDM-estimated breast volume. A previous study involving 110 NSM patients
      • Georgiou C.A.
      • Ihrai T.
      • Chamorey E.
      • Flipo B.
      • Chignon-Sicard B.
      A formula for implant volume choice in breast reconstruction after nipple sparing mastectomy.
      examined the clinical data (e.g., mastectomy weight) associated with the implant volume and provided a mathematical equation for implant the volume, with a correlation coefficient of 0.66. These correlation coefficients and R2 values are lower than those in the main results, suggesting a lower prediction performance. Although the previous study
      • Georgiou C.A.
      • Ihrai T.
      • Chamorey E.
      • Flipo B.
      • Chignon-Sicard B.
      A formula for implant volume choice in breast reconstruction after nipple sparing mastectomy.
      did not provide details regarding implant types, the type of implant used likely varied, as in the present study (six implant types, initially). Our experience demonstrated that it is difficult to develop a universal prediction model to fit various implant types. Given the implant type and category distribution in the present data set, a target patient group was selected (NSM patients with a Mentor MP/MPP/HP breast implant) and the prediction performance was significantly increased. This reveals the importance of clustering NSM patients by implant choice based on an appropriate and common feature to decrease the heterogeneity.
      In the present study, the mean implant volume and specimen weight across all patients was 264.31 cc and 314.06 g, respectively, which may raise concerns that these two formulas would not generalize to Western women. It is well known that Asian women have lower breast volume and BMI compared to those in Western women. However, the implant volume in present study had a large range (from 150 to 600 cc), and there were 64 patients (35.16%) requiring an implant larger than 300 cc. In addition, the specimen weight of most of the NSM patients in two previous NSM-related studies conducted in France
      • Georgiou C.A.
      • Ihrai T.
      • Chamorey E.
      • Flipo B.
      • Chignon-Sicard B.
      A formula for implant volume choice in breast reconstruction after nipple sparing mastectomy.
      and the United States
      • Dull B.
      • Conant L.
      • Myckatyn T.
      • et al.
      Nipple-sparing mastectomies: clinical outcomes from a single academic institution.
      ranged from 125 to 625 g. Given the specimen weights and implant volumes in most of the patients in the present study, the obtained formulas were shown to have good performance and a small allowance in estimating the implant size when the specimen weight was less than 700 g. Moreover, to avoid potential askewness or movement of the implant after reconstruction, the transverse rectus abdominis myocutaneous flap reconstruction procedure is preferred for Asian women requiring a larger implant volume (>500 cc). Therefore, these two formulas are likely suitable for cases of immediate implant breast reconstruction with an implant size of less than 500 cc.
      The implant volume estimation formulas were based on our past experience with implants. Determining whether the chosen implant volume was the most appropriate for the patient is an important issue. In Figure 2, some data points appearing in the left and upper sides were significant outliers. These data points indicate that some patients have a smaller specimen weight but accepted a larger implant volume during the breast reconstruction after NSM. This reflects the subjective nature of esthetics based on patient preferences and can result in asymmetry or implants that were smaller or larger than the actual breast size (outliers), based on past experience. Given that esthetic preferences are subjective, the possible reasons for the inconsistency in size may include a congenital asymmetry in breast size or increased BMI. In such cases, physicians may need to choose a larger implant size to balance the breast size and maintain esthetics.
      In the curve-fitting procedure, the formulas were fully adjusted (including the use of a robust regression scheme to decrease the effect of outliers) in an attempt to obtain higher correlations with the implant volume. Currently, the formulas cannot aid in decisions regarding the appearance differences among Mentor MP/MPP/HP implants. However, adding the consideration of the appearance decision would improve the utility of the formulas, and this will be addressed in future work.
      To provide an assessment of the postoperative outcomes using the formula based on the specimen weight that was developed in the present study, a preliminary prospective study was performed. Twenty-three patients were enrolled as a prospective cohort from January 2016. As breast volume data were obtained before December 2015, the breast volume data in the prospective cohort was limited, and only the formula using the specimen weight was evaluated. Specimen weights were used to estimate the implant size based on the obtained formula; the resulting SD was 57.73 and the standard error of the mean was 12.04. These results are very similar to those for the main retrospective data set, and two participating surgeons (S-L.L. and D-R.C.) confirm the satisfactory outcomes of this preliminary study.
      These implant volume estimate formulas benefit surgeons by providing a preoperative implant volume assessment using the breast volume and an intraoperative assessment using the specimen weight. Because of limitations in collecting breast volume data (for some patients, only the presentation file was maintained and the loss of the raw file resulted in an inability to generate the volumetric data), the volumetric data set was small and could not cover most of the patients; therefore, a selection bias during the formula development was unavoidable. Considering that the retrospective data set was relatively small and the FFDM-based formula has not been fully tested, the generalizability of these two formulas still requires more testing, especially in different populations. However, given the relatively high R2 value in the main retrospective analysis and the relatively small allowance obtained in the prospective preliminary study results, we believe that the results of the present study are acceptable despite these limitations.

      Conclusions

      Appropriate implant choices are highly related to the esthetic results, yet supporting studies are lacking and much depends on the physician's experience. In the present study, several estimation procedures were evaluated, and formulas based on the specimen weight and breast volume of real NSM patients were obtained to help physicians estimate and select the appropriate implant volume in DTI reconstruction. The implant size estimation formulas obtained in the present study may be applied in a majority of patients. However, the final decision regarding the implant volume must be taken by experienced physicians.

      Acknowledgment

      All authors would like to thank Volpara Solutions for their support in this research plan and for providing the right to use Volpara software. This work was supported by the Department of Research, Changhua Christian Hospital, Changhua, Taiwan (No. 106-CCH-IRP-006 and 106-CCH-MST-127).
      Authors' contributions: W-C.S. and H-J.Y. were involved in article drafting, method design, statistical analysis, and curve-fitting procedures. D-R.C. was involved in framing the main propose of the study and NSM patient data collection. H-K.W. was involved in acquiring the FFDM data. D-R.C., S-L.L., and H-W.L. were involved in patient recruitment. Y-L.H. was involved in the mammography analysis. D-R.C. and S-L.L. were involved in the prospective study test.

      Disclosure

      None of the authors have any commercial interest or sources of financial or material support to declare.

      References

        • Rusby J.
        • Smith B.
        • Gui G.
        Nipple-sparing mastectomy.
        Br J Surg. 2010; 97: 305-316
        • de Alcantara Filho P.
        • Capko D.
        • Barry J.M.
        • et al.
        Nipple-sparing mastectomy for breast cancer and risk-reducing surgery: the Memorial Sloan-Kettering Cancer Center experience.
        Ann Surg Oncol. 2011; 18: 3117
        • Leibman J.A.
        • Styblo T.M.
        • Bostwick III, J.
        Mammography of the postreconstruction breast.
        Plast Reconstr Surg. 1997; 99: 698-704
        • Yip J.M.
        • Mouratova N.
        • Jeffery R.M.
        • et al.
        Accurate assessment of breast volume: a study comparing the volumetric gold standard (direct water displacement measurement of mastectomy specimen) with a 3D laser scanning technique.
        Ann Plast Surg. 2012; 68: 135-141
        • Highnam R.
        • Brady M.
        • Yaffe M.J.
        • Karssemeijer N.
        • Harvey J.
        Robust breast composition measurement-VolparaTM.
        in: Digital Mammography. IWDM 2010. Lecture Notes in Computer Science. 6136. Springer, Berlin, Heidelberg2010: 342-349
        • Tromans C.
        Comparing Personalized Mean Glandular Dose Estimates Between X-Ray Systems Over Time in Mammography.
        (European Congress of Radiology 2014)2013https://doi.org/10.1594/ecr2014/C-0360
        • Bassett Jr., G.
        • Koenker R.
        Asymptotic theory of least absolute error regression.
        J Am Stat Assoc. 1978; 73: 618-622
        • Heiser W.J.
        Correspondence analysis with least absolute residuals.
        Comput Stat Data Anal. 1987; 5: 337-356
        • Georgiou C.A.
        • Ihrai T.
        • Chamorey E.
        • Flipo B.
        • Chignon-Sicard B.
        A formula for implant volume choice in breast reconstruction after nipple sparing mastectomy.
        Breast. 2012; 21: 781-782
        • Dull B.
        • Conant L.
        • Myckatyn T.
        • et al.
        Nipple-sparing mastectomies: clinical outcomes from a single academic institution.
        Mol Clin Oncol. 2017; 6: 737-742