Page 264
12
Silicone Implants and Breast Imaging
Women with silicone breast implants undergo imaging evaluations to detect early breast cancer just as women without breast implants. In addition, women undergo imaging to assess the integrity of these implants. Several characteristics of implants and the techniques of their placement affect imaging evaluations. These include the major types of prostheses (e.g., silicone elastomer shells and fillers of silicone gel, saline, both gel and saline mixed, or multiple lumens), the great variety of implants of current and historical manufacturers (described in Chapter 3), and the different placement of implants (subcutaneous), subglandular, or submammary, above the chest wall muscles, and submuscular or subpectoral under those muscles, as described in Chapter 5).
Implant Integrity Assessment
Implant rupture is defined in Chapter 5 as silicone gel detectable on the outer implant and/or capsular surface. This does not necessarily imply a complete disruption of the implant shell, but indicates only loss of shell integrity and movement of silicone gel outside the elastomer shell. As discussed earlier, rupture, which includes what some have termed leakage, is different from gel fluid diffusion, which refers to the diffusion of the lower molecular weight silicone fluid that permeates the silicone gel through the implant shell into the capsule or surrounding tissues.
Breast implants are encased by a fibrous capsule, and a potential or actual fluid-filled space is produced between the implant shell and the
Page 265
surrounding capsule, as described in Chapters 3 and 5. Rupture involving only the silicone shell, with free silicone gel still contained by the surrounding fibrous capsule, are defined as intracapsular rupture. Disruption of both the implant shell and fibrous capsule allows silicone access to breast tissue and is defined as an extracapsular rupture. Normal silicone fluid diffusion is detectable only rarely by imaging examinations.
Other terms associated with loss of shell or fibrous capsule integrity are also in common usage. Herniation indicates focal bulging of an intact implant through a defect in the surrounding fibrous capsule. Extrusion implies a sudden flow of silicone gel through defects in the implant shell and fibrous capsule, which may occur with traumatic events. Infiltration is a slow movement of extracapsular silicone gel into surrounding breast or other tissue. Extravasation is an inclusive term encompassing extrusion and infiltration, whereas migration refers to extracapsular silicone gel movement away from the implant.
The frequency of implant rupture is unknown. Chapter 5 discusses the reasons for this, which include such factors as the changing composition of implants, the decades-long observation required in some cases, the study of nonrepresentative groups of women with implants, and incomplete or imprecise detection of rupture. Some confusion is also occasionally caused by the separation of rupture into leakage and rupture or disruption categories. Estimates of implant rupture prevalence range from 0.3 to 77%, as reported earlier. Rupture prevalence depends at least in part on implant characteristics such as elastomer shell thickness and strength; thus descriptions of rupture prevalence must consider and identify the types and ''generations" of implants. Breast implant integrity can be evaluated clinically; mammographically; and with computed tomography, ultrasound, and magnetic resonance imaging (MRI).
Mammography
The intact silicone gel-filled implant appears as a radiodense structure sharply circumscribed from surrounding breast tissue. At times, the implant fibrous capsule may be visible just superficially to the implant shell. Dystrophic calcification can be identified in the fibrous capsule (Benjamin and Guy, 1977). Calcification is often seen and could represent a long-term inflammatory response to the breast prosthesis (Barker et al., 1978; Cocke et al., 1985; Ginsbach et al., 1979; Inoue et al., 1983; Koide and Katayama, 1979; Peters et al., 1995d; Redfern et al., 1977; Young et al., 1989). Textured implants may disorganize capsular fibrotic reactions and decrease capsular contracture. Use of textured implants could be a source of a false-positive mammographic diagnosis of rupture. The presence of a textured implant is easily detected by its brush border (Piccoli, 1968). The
Page 266
capsule around a textured implant grows into and around the pores and small papillary projections of the shell surface. Silicone can be observed in these irregularities, and this may mimic implant rupture (Kasper, 1998).
Direct mammographic evidence of rupture is related to demonstration of extravasated silicone, silicone droplets, or calcified silicone in surrounding breast tissue as a result of an extracapsular rupture. Because of the radiodensity of silicone, direct evidence of intracapsular rupture contained by the intact fibrous capsule may not be possible. Indirect mammographic signs of intracapsular rupture include changed implant dimensions compared to prior studies, an ill-defined border or irregular implant density, marked bulging of an implant border, and silicone in mammary gland ducts or lymphatics (Destouet et al., 1992; Ganott et al., 1992; Gorczyca et al., 1994a; Peters et al., 1995d; Reynolds et al., 1994; Theophelis and Stevensen, 1986).
Reporting of the sensitivity and specificity of detection of implant rupture varies depending on the way rupture is defined mammographically. Accurate sensitivity and specificity also depend on an accurate verification of rupture. As discussed in Chapter 5, careful direct visual examination of a breast implant on explantation is considered the most reliable standard. In the reports reviewed in Chapter 5, this standard was generally (but not always; see, for example, Peters and Pugash, 1993, Table I) used to determine the performance of imaging technologies in diagnosing rupture (i.e., their sensitivity and specificity). Even the explantation standard is occasionally imperfect, however, because intact implants may inadvertently and unknowingly be torn during explantation, leading to a mistaken diagnosis of rupture, or tiny holes that allow gel through the shell may not be seen, leading to a mistaken diagnosis that the implant is intact.
Handel et al. (1992) described two levels of concern in the mammographic detection of ruptures. Suspicious findings included major bulges, anteroposterior flattening, and irregular, ill-defined margins of the implant, while silicone outside the implant was considered diagnostic. If signs considered diagnostic defined a mammogram as a true positive for rupture, the sensitivity was only 15% but specificity was 100%. However, if both suspicious and diagnostic findings were included, then sensitivity and specificity were 69 and 82%, respectively (Handel et al., 1992). Everson et al. (1994) demonstrated a 23% mammographic sensitivity with a 98% specificity. These authors also felt that mammography was specific for the detection of extracapsular implant rupture but was of limited use for intracapsular rupture. Ahn et al. (1994a) noted an 11% sensitivity and 89% specificity. They reported a high false-negative rate of 33.3%. These authors also confirmed the ability of mammography to detect free silicone, indicating an extracapsular rupture, but found that free silicone in the
Page 267
posterior aspect and intrapectoralis muscular areas was difficult to identify (Ahn et al., 1994a). Andersen et al. (1989) also found that mammograms were excellent for detecting free silicone in the breast, demonstrating an accuracy of 90%. However, if a mammogram with any findings (in addition to free silicone) suggestive of rupture was called positive, the sensitivity fell to 67% (Andersen et al., 1989). Other authors have reported varying sensitivities in series and reviews of various size, for example, 89% (Cohen et al., 1997); 62% (Nemecek, cited in Samuels et al., 1995); 16.2% (Robinson et al., 1995); and 45%-67% (Samuels et al., 1995).
Ultrasound
Ultrasound is used routinely as an adjunct to film mammography for evaluating palpable breast masses or clinically occult masses detected by x-ray. Ultrasound evaluation of silicone gel implants can be a sensitive technique for the evaluation of implant integrity. Since the examination is highly operator dependent, sensitivity and specificity are more variable than with other imaging techniques and depend on the experience of the radiologist in performing and interpreting the examination and on the scanning techniques utilized. A high-resolution linear array transducer of 7-10 MHz should be used, and real-time scanning by the interpreting physician is strongly recommended to eliminate confusion due to artifacts that commonly occur (Gorczyca et al., 1994b). Artifactual echoes that are anteriorly positioned in the silicone gel implant, as well as echoes produced behind the implant in the chest wall and pictured within the lumen of the implant, may be confused with implant rupture. These are effects of the varying speed of sound in silicone gel and breast tissue. If the operator and interpreter are inexperienced, these artifacts can be confused with findings of loss of implant integrity (Forsberg et al., 1996).
There are several sonographic signs relating to the evaluation of implant integrity. The interface of the breast with the implant may contain highly echogenic signals obscuring posterior breast tissue. This has been termed the "snowstorm" sign and represents free silicone gel adjacent to the implant (Harris et al., 1993). The presence of hypoechoic nodules that may be associated with this snowstorm sign is thought to represent larger conglomerates of silicone within breast tissue (Herzog, 1989; Van Wingerden and Van Staden, 1989) and also to indicate an extracapsular rupture. Echogenic material within the implant lumen (Berg et al., 1993; DeBruhl et al., 1993) as well as the presence of either single, multiple and continuous, multiple and discontinuous, or parallel ("stepladder" sign) echogenic lines or bands may correspond to a disrupted and collapsed implant shell contained within an intact fibrous capsule, i.e., an intracapsular rupture (DeBruhl et al., 1993; Gorczyca et al., 1992).
Page 268
A wide range in the sensitivity and specificity of ultrasound detection of rupture is reported in the literature. The sensitivities range from 25 to 100%, and specificities range from 50 to 99% (Ahn et al., 1994a; Berg et al., 1995; Caskey et al., 1994; Chilcote et al., 1994; Chung et al., 1996; DeBruhl et al., 1993; Everson et al., 1994; Harris et al., 1993; Liston et al., 1994; Medot et al., 1997; Palmon et al., 1994; Peters and Pugash, 1993; Petro et al., 1994; Reynolds et al., 1994; Venta et al., 1996). However, the majority of these studies report a sensitivity of 48-74% (Ahn et al., 1994a; Berg et al., 1995; Caskey et al., 1994; Chung et al., 1996; DeBruhl et al., 1993; Everson et al., 1994; Harris et al., 1993; Palmon et al., 1994; Peters and Pugash, 1993; Reynolds et al., 1994; Venta et al., 1996) and a specificity of 50-90% (Berg et al., 1995; Caskey et al., 1994; Chilcote et al., 1994; Chung et al., 1996; Everson et al., 1994; Medot et al., 1997; Palmon et al., 1994; Reynolds et al., 1994; Venta et al., 1996). A low sensitivity reported in one article (Chilcote et al., 1994) was associated with separation of data into ruptures and all implant failures. The ruptures were defined as tears greater than 10 cm in diameter. When this definition was used, sensitivity was 50%. When rents less than 0.5 cm were included in the definition, sensitivity dropped to 25%. The article reporting 100% sensitivity used both wall redundancy and "atypical silicone," neither of which was described. However, the sensitivity is based on only seven ruptured implants (Petro et al., 1994). The study reporting a 99% specificity (Peters and Pugash, 1993) was a retrospective study of 150 consecutive office patients with only 8 going on to surgery. Diagnoses were based on a combination of ultrasound and clinical analysis; one woman with abnormal ultrasound turned out to have intact implants at surgery.
Although sensitivity and specificity show wide variation, it appears that ultrasound is more specific than it is sensitive. Several authors calculated the sensitivity and specificity of the various ultrasound signs of altered implant integrity. Caskey et al. (1994) demonstrated that low-level echoes within the implant were the most sensitive sign of rupture (55%), while adjacent echogenic noise had the highest specificity (97%). Echogenic noise surrounding an implant had a 97% correlation with rupture, but this sign was seen in only 5% of ruptures. DeBruhl et al. (1993) found the presence of internal parallel echogenic lines to be the most sensitive sign, occurring in 70% of ruptured implants. Palmon et al. (1994), however, demonstrated that linear echoes can be seen in most silicone breast implants, and their presence or absence is not useful in predicting rupture. Medot et al. (1997) analyzed the detection of implant integrity with ultrasound based on the presence or absence of capsular contracture. The sensitivity and specificity for rupture was 41 and 70%, respectively, with capsular contracture, and 69 and 74%, respectively, without capsular con-
Page 269
tracture. They concluded that ultrasound is reliable only for indicating rupture in the absence of capsular contracture.
Magnetic Resonance Imaging
Silicone gel can be differentiated from breast tissue by taking advantage of its unique nuclear magnetic resonance properties. The basis of magnetic resonance imaging and chemical characterization is detection of a resonance signal from perturbed protons within the body. Since protons are the most abundant nuclear elements, a detectable resonance signal can be expected. Protons perturbed by radiofrequency waves in a homogeneous magnetic field emit a second radiofrequency wave when the stimulating frequency is discontinued. The properties of this emitted resonance frequency depend on the molecule to which the protons are attached and the nature of the disturbing input radiofrequency. Through a judicious use of sequence parameters, the protons in breast tissue (i.e., in water) and those in silicone can be separated (Derby et al., 1993).
Although a detailed analysis of the various MRI sequences is not within the scope of this report, a brief description of basic principles will improve understanding of MRI examinations of women with breast implants. T1 and T2 imaging refer to imaging sequences that can enhance signal from protons in various biochemical compounds. Silicone, on a T1-type image, will have a low signal and, on a T2-type image, a high signal. Protons associated with fat will also have a high signal on T2 images, although lower than silicone. There are several techniques that are successful in separating proton resonance signals emanating from fat, water, and silicone. The three-point Dixon technique (Schneider and Chan, 1993) produces an image that demonstrates a signal only from silicone and nullifies signals from fat and water. The RODEO (rotating delivery of excitation off-resonance) technique can selectively eliminate signals within the proton resonance range of silicone (Harms et al., 1995), allowing silicone detection by determining which signals are lost after the RODEO technique is applied.
The accuracy of MRI in evaluating the integrity of breast implants derives from its ability to detect rupture of the implant shell contained by an intact fibrous capsule, that is, intracapsular rupture. Several signs of rupture have been described with MRI. The "linguine sign" (Gorczyca et al., 1994c) is a series of low-signal curvilinear lines within the high-signal silicone gel and represents segments of free-floating implant shell. In instances of intracapsular rupture without complete collapse of the implant shell, silicone is present both within and outside the shell. This silicone may invaginate a portion of the implant shell producing the so-called keyhole, teardrop, or noose sign (Berg et al., 1994, 1995; Gorczyca, 1994c;
Page 270
Mund et al., 1993). The "subcapsular line" sign is also an indication of a ruptured implant without complete disruption of the implant shell. Silicone completely surrounds the shell and highlights the implant envelope as a low-signal line subtended by intra- and extracapsular silicone gel closely paralleling the intact fibrous capsule (Benjamin and Guy, 1977).
As noted earlier, the various types of breast implants and surgical procedures must be considered to avoid false-positive diagnoses of implant rupture. The most prevalent pitfall is the presence of radial folds, which are normal invaginations of the periphery of the implant shell. These folds appear thicker than the low-signal lines seen with shell disruption since they represent two adjacent portions of the implant shell. They rarely traverse the implant on all images and when imaged in plane may present a sheetlike appearance that is unusual in true implant shell disruption (Benjamin and Guy, 1977). Double-lumen devices have an inner shell containing silicone gel and a surrounding compartment of saline. The outer saline-containing shell may rupture, resulting in the image of a normal single-lumen implant. This can be a source of confusion when placement of a double-lumen implant was recorded. At times, an intact single-lumen implant surrounded by reactive fluid contained by the fibrous capsule can be mistaken for a normal double-lumen device. Water droplets within the silicone gel can be detected by MRI and may also indicate loss of implant shell integrity. This finding may be deceptive since saline or saline with antibiotics or steroids may have been injected into the implant at the time of placement (Berg et al., 1994).
The sensitivity and specificity of MRI for implant rupture is greater than that of mammography or ultrasound. This is especially true when magnetic resonance coils specifically designed for the breast are utilized. Soo et al. (1997) reported a sensitivity of 88% utilizing the linguine sign and any two signs indicative of silicone on the surface of the implant (keyhole, subcapsular, or teardrop). The specificity of the examination was 92%, with a positive predictive value (PPV) of 43% and a negative predictive value (NPV) of 85% (Soo et al., 1997). Gorczyca et al. (1992) demonstrated a 76% sensitivity and 97% specificity utilizing a shoulder coil, and in 1994, they demonstrated an 89% sensitivity and 97% specificity using a dedicated breast coil (Gorczyca et al., 1994a). Several studies report a low sensitivity of MRI for detection of implant integrity. Weizer et al. (1995) described a sensitivity of 46%; the authors state that the use of a body coil was a major reason for the low sensitivity. Reynolds et al. (1994) used a shoulder coil for most of their implant examinations and recorded a sensitivity of 69%; this study involved only 13 patients and 24 implants. Ahn et al. (1993) demonstrated a 76% sensitivity and Middleton (1998b) a 74% sensitivity. All of the examinations in the former study were done with a body or surface coil not dedicated for breast imaging,
Page 271
and the latter study used a mixture of dedicated and nondedicated coils. Modern MRI scanning is a highly sensitive and specific test for the detection of implant ruptures with the use of proper equipment and imaging parameters.
There are advantages and disadvantages to all imaging modalities used in the evaluation of implant integrity. Mammography is rapid and inexpensive, but it is very inaccurate for intracapsular rupture and will reliably detect extruded silicone only in an extracapsular rupture. Ultrasound has the potential to detect both intra- and extracapsular ruptures, is inexpensive, and uses no ionizing radiation. However, it is a highly operator-dependent study that is less accurate in the presence of capsular contracture and is unable to visualize the posterior surface of the implant. The sensitivity of the examination is greater than mammography, however. MRI has the highest sensitivity and specificity for evaluation of implant integrity and has none of the limitations of mammography or ultrasound, but it is expensive and time-consuming. MRI requires modern dedicated breast-imaging coils, and necessitates a thorough knowledge of implants and of potential diagnostic pitfalls. When these conditions are present, the interoperator agreement may be excellent, however (Brown et al., 1999).
Although the sequence of modalities used to evaluate implant integrity will vary, the committee has concluded based on current information, that the following steps are reasonable. A clinical suspicion of loss of implant integrity such as a sudden size change, pain, or asymmetry should initiate the imaging sequence. Mammography and ultrasonography can be the initial imaging examinations. If both are normal, the clinician may wish to follow the patient. If these examinations are unequivocally abnormal, then explantation may be considered on a case-by-case basis. Any equivocal or suspicious result of mammography or ultrasound study should be followed by MRI evaluation. If this examination is positive for rupture, explantation may be considered. Some have suggested MRI on all clinically suspicious cases or for routine examination of women with implants and clinical signs suggestive of loss of implant integrity (Samuels et al., 1995). The committee finds that there is no convincing evidence to support routine screening of asymptomatic women for implant rupture. Data on the cost/benefit of routine screening for rupture are lacking, however. Studies that provided such data and analysis might allow a firm conclusion on whether screening for rupture is indicated or not indicated in asymptomatic women with silicone breast implants, either as a routine procedure, or in specific situations such as women with rupture prone implants or in circumstances of changing technologies or certain clinical comorbidities. To justify routine screening of the general population of asymptomatic women with silicone breast implants, such studies would
Page 272
have to provide convincing scientific data showing that routine screening and a consequent intervention effectively reduced complications and morbidity secondary to implant rupture. Table 12-1 reviews reports on the evaluation of implant integrity.
Mammography and Implants
Technique
Modern mammography is divided into screening and diagnostic examinations. Screening mammography is an x-ray examination to detect unsuspected breast cancer at an early stage in asymptomatic women. It usually consists of two views of each breast, a mediolateral oblique (MLO) and a craniocaudal (CC) view. The examination is performed by a qualified technologist, often in the absence of the interpreting physician. Diagnostic mammography is an x-ray examination to evaluate abnormal physical findings in the breast or abnormal findings detected with a screening mammogram. Diagnostic mammography is performed under the on-site supervision of a qualified interpreting physician.
The presence of silicone gel-filled breast implants may interfere with standard mammography since silicone is radiopaque, and the physical presence of the implant compresses fat and glandular tissues, creating more homogeneous dense tissue that frequently lacks the contrast needed to detect subtle early features associated with breast cancer. Eklund et al. (1988) described a modified compression technique that permits more effective imaging of breasts with implants. The prosthesis is displaced posteriorly and superiorly against the chest wall while the breast tissue is gently pulled anteriorly onto the image receptor and held in place by the compression device. This maneuver should be used for both the CC and the MLO views, and such views are termed implant-displaced views. Breast implants are surrounded by a fibrous capsule that may be soft or hard. If the capsule is hard and immobile, it may be impossible to perform the implant-displaced views. The MLO view may be replaced by the 90-degree lateral view if the latter depicts more breast tissure in individual patients. When there is clinical concern for lesions cephalad to the implant between the 11 and 1 o'clock positions or caudad to the implant between the 5 and 7 o'clock positions, the 90 degree lateral view can be helpful (Heinlein and Bassett, 1997). Thus, the current standard for mammography of women with implants is both a nondisplaced and an implant-displaced view for each of the routine views. This examination results in four views per breast, the CC and MLO views in both the implant-displaced and the standard modes.
The augmented breast presents unique imaging problems depending
Page 273
on the experience and expertise of the technologist and interpreting physician, surgical techniques, and characteristics of the implant. Even with the modified technique described above, the amount of breast tissue visualized will be limited by the implant. Discussion of the utility of mammography with implants began as early as 1968 when Mendes-Filho and Ludovici advocated preimplant mammography to facilitate later comparison. In 1974, Rintala and Svinhufvud concluded that the prosthesis did not hamper the technical performance of the examination. Cohen et al. (1977) disagreed, concluding that gel implants did obscure portions of the breast. In one of the first actual estimates, Wolfe (1978) reported breast tissue nonvisibility in the presence of silicone gel implants at about 25%. Hayes et al. (1988) examined six sets of mammograms in the CC and MLO views and subtracted the calculated volume of the implant from the calculated volume of both the implant and the visualized tissue. They found that the proportion of obscured glandular tissue ranged from 15 to 100%, with an average of 41%. Silverstein et al. (1990a) calculated the area of mammographically visualized breast tissue before and after augmentation mammaplasty using a transparent grid to measure surface area. Women whose implants were placed in a subglandular position had a mean decrease in measurable tissue of 49% with nondisplaced mammography and 39% with implant-displaced views. The decrease in measurable tissue was 28 and 9% in nondisplaced and displaced views, respectively, in patients with subpectoral implants (Silverstein et al., 1990a). More qualitatively, Destouet et al. (1992) rated the quality of implant-displaced views from excellent to poor based on an estimate of improved breast tissue visualization with these views compared to standard views in 252 women. They found the subpectoral position of the implant provided implant-displaced views uniformly rated as excellent, compared to only 7% so rated for the subglandular position (Destouet et al., 1992). Jensen and Mackey (1985) also found part of the breast obscured to mammography by implants. Lindbichler et al. (1996), using qualitative assessments rating mammograms as good, acceptable, and limited, found that examinations in 29% of subpectoral and 22% of subglandular silicone gel implanted breasts were of limited quality.
The finding of reduced visualization in mammograms of women with breast implants has spurred interest in utilizing radiolucent implant fillers. Young et al. (1993a) tested the biocompatibility of peanut oil (a medium-chain triglyceride) and also tested the radiographic properties of oil-filled versus saline-filled prostheses implanted in rabbits. With silicone gel as a control, 10 ml of sterile peanut oil was injected into rats. These rats demonstrated a rapid absorption of peanut oil with no abnormalities at histologic evaluation of the lungs, liver, kidneys, and tissues adjacent to the injection sites. Radiography of the two types of implants
Page 274
TABLE 12-1 | |||
Study | No. (patients/breasts) | Study Design | Technique |
Soo et al., 1996 | 37/72 | Surgical proof,a retrospective | MRI |
Soo et al., 1997 | 44/86 | Surgical proof, retrospective | MRI |
Gorczyca et al., 1992 | 70/140 | Surgical proof, retrospective | MRI |
Gorczyca et al., 1994 | 41/81 | Surgical proof, retrospective | MRI |
Monticciolo et al., 1994 | 23/38 | Surgical proof, prospective | MRI |
Quinn et al., 1996 | 54/108 | Surgical proof, prospective/retrospective | MRI/ |
Ahn et al., 1993 | 80/ | Surgical proof, retrospective | MRI |
Andersen et al., 1989 | 24/ | Surgical proof, retrospective | Mammography |
Chung et al., 1996 | 98/192 | Surgical proof, prospective | US |
Caskey et a1., 1994 | 31/59 | Surgical proof, prospective | US |
Chilcote et al., 1994 | 25/42 | Surgical proof prospective | US |
Medot et al., 1997 | 65/122 | Surgical proof, retrospective | US |
DeBruhl et al., 1993 | 28/57 | Surgical proof, prospective | US |
Venta et al., 1996 | 43/78 | Surgical proof, prospective | US |
Harris et al., 1993 | 22/29 | Medical records, surgical proof | US |
Palmon et al., 1997 | 33/64 | Prospective, evaluation only sign of echo lines in implant | USthick lines |
Ahn et al., 1994a | 29/59 | Surgical proof, prospective | US |
(Table continued on next page)
Page 275
(Table continued from previous page)
Study | Sensitivity | Specificity | PPV (%) | NPV (%) | Accuracy | |
Soo et al., 1996 | TD48% of rupture | 0 | 0 | 0 | 0 | |
Linguine48% of rupture | ||||||
Folds4% of rupture | ||||||
Soo et al., 1997 | Linguine44% | 100% | 100 | 58 | 0 | |
Subcap, L, or KS, 94% | 87% | 90 | 92 | |||
L or any two of SC, KS | ||||||
TD, 88% | 92% | 93 | 85 | |||
Gorczyca et al., 1992 | 76% | 97% | 84 | 96 | 94 | |
Gorczyca et al., 1994 | 3-point Dixon, 60% | 3-point | 0 | 0 | 0 | |
Dixon, 97% | ||||||
FSE, 89% | FSE, 97% | |||||
Monticciolo et al., 1994 | 94% | 100% | 0 | 0 | 0 | |
81% | 100% | |||||
Quinn et al., 1996 | 87% | 78% | 0 | 0 | 0 | |
93% | 92% | 0 | 0 | 0 | ||
Ahn et al., 1993 | 76% | 97% | 0 | 0 | 0 | |
Anderson et al., 1989 | 67% | 0 | 0 | 0 | 0 | |
Chung et al., 1996 | 74% | 89% | 77 | 88 | 0 | |
Caskey et al., 1994 | 55% | 84% | 0 | 0 | 0 | |
Chilcote et al., 1994 | All implant failures, 25% | 75% | 0 | 0 | 54 | |
Rupture only, 50% | 75% | 71 | ||||
Medot et al., 1997 | With CC, 41.2% | 70% | 53.9 | 58.3 | 0 | |
Without CC, 68.7% | 73.6% | 61.1 | 79.6 | 0 | ||
DeBruhl et al., 1993 | 70% | 92% | 82 | 85 | 0 | |
Venta et al., 1996 | 50% | 55% | 58 | 91 | 0 | |
Harris et al., 1993 | 65% | 0 | 0 | 0 | 0 | |
45% | ||||||
50% | ||||||
Palmon et al., 1997 | 48% | 50% | 0 | 0 | 0 | |
48% | 50% | |||||
43% | 50% | |||||
70% | 92% | 0 | 0 | 0 | ||
11% | 89% | |||||
81% | 92% |
Continued
Page 276
TABLE 12-1 | |||
Study | No. (patients/breasts) | Study Design | Technique |
Petro et al., 1994 | 0/22 | Surgical proof, prospective | US, wall redundancy |
Middleton, 1998b | 877/1626 | Surgical proof, retrospective | MRI |
Dobke et al., 1994 | 39/74 | Surgical proof, retrospective | MRI |
Berg et al., 1995 | /144 | Surgical proof, prospective | MRIrupture |
Weizer et al., 1995 | 81/160 | Surgical proof, prospective | US |
Reynolds et al., 1994 | 13/24 | Surgical proof | Mammography |
Peters and Pugash, 1993 | 150/ | Some surgical, some clinical, retrospective | US |
Ahn et al., 1994b | /22 | Surgical proof, retrospective | CT |
NOTE: FSE = fast spin echo; KS = keyhole sign; L = linguine; TD = teardrop; and US = ultrasound. | |||
aImaging verified at explant. |
(Table continued on next page)
placed in 21 rabbits showed the peanut oil-containing implants to be radiolucent while those filled with saline obscured surrounding soft tissue (Young et al., 1993a). Using an American College of Radiology (ACR) phantom to simulate microcalcifications and soft tissue masses, Gumucio et al. (1989) placed various implant fillers over the ACR phantom and then evaluated the phantom scores produced with each. The shells filled with silicone gel or a mixture of saline and silicone gel completely obscured the phantom. Shells filled with gelatin allowed limited visibility of the simulated calcifications but obscured the masses. Peanut oil and sunflower oil allowed the best resolution of the artifacts within the ACR phantom.
Beisang and Geise (1991) reported polyvinylpyrrolidone (PVP), known as Bio-Oncotic gel, as a filler whose radiographic properties were close to normal breast tissue. Young et al. (1993a) found peanut oil to allow imaging of the ACR phantom most clearly; the oil was four times
Page 277
(Table continued from previous page)
Study | Sensitivity | Specificity | PPV (%) | NPV (%) | Accuracy |
Petro et al., 1994 | 81% | 71% | |||
29% | 87% | 0 | 0 | 0 | |
100% | 71% | ||||
Middleton, 1998b | 74% | 98% | 0 | 0 | 0 |
Dobke et al., 1994 | 100% | 91% | 0 | 0 | 0 |
Berg et al., 1995 | 98% | 91% | |||
50% | |||||
65% | |||||
25% | 57% | 0 | 0 | 0 | |
Weizer et al., 1995 | 47% | 83% | 55 | 83 | 0 |
46% | 88% | 60 | 83 | ||
Reynolds et al., 1994 | 69% | 82% | 0 | 0 | 0 |
54% | 64% | ||||
69% | 55% | ||||
Peters and Pugash, 1993 | 67% | 99% | 0 | 0 | 0 |
Ahn et al., 1994b | 94% | 100% | 0 | 0 | 0 |
more radiolucent than saline or Bio-Oncotic gel and about 45 times more radiolucent than silicone gel. In a unique study conducted by Handel et al. (1993b), a patient scheduled to undergo a mastectomy consented to a series of CC views utilizing implant shells filled with various filler materials interposed between the breast and the film cassette. Lesions were best seen through the triglyceride solution. Lesions were observed best with Bio-Oncotic gel when it was at 10% concentration in saline. No lesions could be visualized through silicone gel (Handel et al., 1993b). Implants of polyester fiber commonly used in vascular grafts were reported not to obscure mammographic detail, although no clinical and little experimental experience with breast implants made of this material is available (Yager and Chaglassian, 1998). The committee has not found enough information to support any conclusions about these and other (e.g., soybean oil) fillers. Bio-Oncotic gel was approved for marketing in the United States for several years about ten years ago, and the new U.S. company,
Page 278
Nova Med, which is introducing a saline implant to the domestic market, markets PVP filled implants in Europe and plans to do so in the United States at some time in the future (see Chapter 3). Soybean oil is under investigation but was recently taken off the market in the United Kingdom because of some reports of adverse reactions.
The adverse publicity regarding silicone implants has caused increasing numbers of women (approaching 40,000 annually in 1994, ASPRS) to have their implants removed. Data on the frequency of implant removal (explantation) are presented in Chapter I of this report. Explantation, at times associated with a mastopexy (breast-lifting procedure), produces mammographic changes that must be recognized. Although imaging findings associated with explantation are not common, architectural distortion from scarring or spiculated masses from residual silicone granulomas can occur and simulate malignancy (Hayes et al., 1993). The presence of bilateral symmetric soft-tissue masses posterior to the glandular tissue, as well as coarse plaque-like calcification from residual calcified capsules left behind, should suggest the possibility of a prior explantation (Peters et al., 1996d; Stewart et al., 1992; Young et al., 1989). At times, residual fibrous capsule remaining after explantation may be particularly thick walled, especially if calcified, and seromas may form within these capsules. They present on mammograms as large, oval, soft-tissue masses with either circumscribed or ill-defined margins (Soo et al., 1995).
Women with breast implants frequently ask whether the compressive forces used during the mammographic examination can cause a loss of implant integrity. There is only rare and anecdotal information in the literature regarding the loss of implant integrity after mammography; this has been discussed in Chapter 5. The subglandular position is the most common location of implants that seem to rupture during mammography. Women usually experience pain and a change in implant shape. In one case, the woman described a ''popping" sensation at the time of compression. However, as noted, this is said to be a rare complication. It is now recommended that the nondisplaced mammographic views use only enough compressive force to maintain a movement-free breast, not the force one would normally use in the nonaugmented breast. The implant-displaced view places no compressive force on the implant but merely displaces the implant posteriorly and superiorly (Beraka, 1995; DeCamara et al., 1993; Hawes, 1990; Williams, 1991). As noted earlier, the committee has concluded that a concern about implant rupture should definitely not discourage properly performed mammography.
Cancer Detection in a Screening Situation
Since the introduction of silicone gel breast prostheses in 1962, about
Page 279
2 million American women have had breast implants. Studies have suggested that breast cancer may be discovered at an advanced stage in women with breast implants and thus have a poorer prognosis. This is ascribed to difficulty in performing mammography which occurs, although to a lesser extent, even with implant-displaced views. The efficacy of mammography in significantly decreasing the mortality from breast cancer has been demonstrated by randomized controlled trials (Kerlikowske, 1997; Smart et al., 1995). The key to the success of mammography is early detection of breast cancer when the tumor is not yet palpable. A review of the cumulative survival by tumor size from the Swedish two-county screening trial undertaken from 1977 to 1985 demonstrates this association. A total of 1,705 cancers were detected in women 40-74 years of age. Fifty-six percent were less than 2 cm in diameter, and the mortality rate from breast cancer in this group was 13% (tumors 1-1.5 cm in diameter are generally not palpable). Tumors 2.0 cm and greater had a mortality rate of 46%, three to fourfold higher (Swedish National Board of Health and Welfare, 1996).
The committee reviewed 12 relevant studies of women with breast implants and cancer detection. This information is summarized in Table 12-2 (Cahan et al., 1995; Carlson et al., 1993; Clark et al., 1993; Dershaw and Chaglassian, 1989; Douglas et al., 1991; Fajardo et al., 1995; Fornage et al., 1994; Grace et al., 1990; Leibman and Kruse, 1993; Schirber et al., 1993; Silverstein et al., 1988, 1992). The accrual dates in these studies, when provided, ranged from 1978 to 1992. Since the majority of the patient accrual took place prior to the establishment of implant-displaced views by Eklund et al. (1988), most of the women studied did not have this maneuver performed. These articles reported 320 women, 278 of whom had mammography with the detection of 264 malignancies. Mammography as the only method of detecting the malignancy occurred in only 38, or 14%, of the women. Palpation as the only method of detection occurred in 126, or 48%, of women, and both mammography and palpation were used in 98, or 37%. (The method of detection was not known in two cases.) Because of the relatively young age of these women, mammography may not have been done routinely and, when done, may have been in response to a clinical indication.
Since it was impossible to determine how many of these women had screening mammography, it is difficult to properly evaluate cancer detection with mammography. Thus, these studies may not represent the detection capabilities of mammography for preclinical disease in women with silicone breast implants. In some of these studies, women with implants had larger primary tumors, more positive axillary nodes, or a lower percentage of palpable tumors visible on mammography than comparison groups of women without breast implants (for example, Carlson et
Page 280
TABLE 12-2 | ||||
Study | Patient | No. of Patients/Mammo | Mammo Only | Palpable Only |
Douglas et al., 1991 | 1978-1988 | 8/6/8 | 1 | 6 |
Schirber et al, 1993 | 6 years (no dates given) | 9/7/9 | 0 | 8 |
Silverstein et al., 1988 | 1981-1989 | 36/36/36 | 1 | 13 |
Fornage et al., 1994 | Retrospective, no dates given | 7/?/22 | 7 | 1 |
Cahan et al., 1995 | 1977-1992 | 22/22/23 | 4 | 17 |
Fajardo et al., 1995 | 1985-1992 | 18/18/18 | 2 | 13 |
Carlson et al., 1993 | No dates given | 35/31/37 | 2 | 16 |
Dershaw and Chaglassian, 1989 | 1984-1987 | 59/59/5 | 1 | 3 |
Silverstein et al., 1992 | 1981-1990 | 42/42/42 | 2 | 19 |
(Table continued on next page)
Page 281
(Table continued from previous page)
Study | Both Mammo and Palpable | ID Views | Cancer Statistics | Implant Location |
Douglas et al., 1991 | 1 | No | Augmented: Mammography with palpable mass, 0% | Subglandular6 |
Nonaugmented: Mammography with palpable mass, 92% | ||||
Schirber et al, 1993 | 1 | ? | Augmented: 4/9 Stage II, 5/9 Stage I | 0 |
Silverstein et al., 1988 | 22 | 32/36 | Augmented: Stage similar to nonmammography detection | Subglandular15 |
Fornage et al., 1994 | ? | "Most no" | Augmented: 8 cysts, 6 cancers, 6 fibroadenomas, 1 granuloma, and 1 fat necrosis | 0 |
Cahan et al., 1995 | ? | ? | Augmented: Mammography without palpable mass, 17% | Subglandular18 |
Fajardo et al., 1995 | 3 | 12/18 | Mammography only | Subglandular17 |
Carlson et al., 1993 | 19 | None | Augmented: | Subglandular29 |
Dershaw and Chaglassian, 1989 | 1 | None | 0 | Subglandular40 |
Silverstein et al., 1992 | 21 | 7 | Augmented: | Subglandular37 |
Continued
Page 282
TABLE 12-2 | ||||
Study | Patient Accrural Dates | No. of Patients/Mammo/Cancers | Mammo Only | Palpable Only |
Clark et al., 1993 | 1982-1991 | 33/29/33 | 8 | 14 |
Leibman and Kruse, 1993 | 1980-1990 | 25/22/25 | 9 | 4 |
Grace et al., 1990 | 1975-1988 | 6/6/6 | 1 | 1 |
NOTE: DCIS = ductal carcinoma in situ; IDC = infiltrating ductal carcinoma; and ILC = infiltrating Iobular carcinoma. |
(Table continued on next page)
al., 1993; Fajardo et al., 1975; Silverstein et al., 1988). Reintgen et al. (1993) in reviewing the effectiveness of mammography in women with breast cancer also supported these findings. Others, however, found no difference in detection, tumor size, or positive axillary nodes between women with or without breast implants (Cahan et al., 1995; Leibman and Kruse, 1993). The committee has found no data on the relative effectiveness of screening for cancer by physical examination of the implanted breast, although descriptions of examination techniques have been reported (Mann, 1995). Studies are needed to determine whether, among women who routinely undergo screening mammography, there are differences in the stage of breast cancers diagnosed in women with and without silicone breast implants.
A study by Deapen et al. (1997) reported cancer staging in a large group of women with implants. This study included 3,182 women who received cosmetic breast implants between 1953 and 1980. In the study group, 31 breast cancer cases were observed compared to 49 expected. The median follow-up was 14.4 years. The authors found that the stage of
Page 283
(Table continued from previous page)
Study | Both Mammo and Palpable | ID Views | Cancer Statistics |
Clark et al., 1993 | 11 | Done post-1988 | Augmented: |
Nonaugmented: | |||
Leibman and Kruse, 1993 | 12 | Done | Augmented:0 |
post-1988 | 18 IDC | ||
Grace et al., 1990 | 4 | None | Augmented: |
breast cancer among women with implants was essentially the same as among women without implants. Women with reconstruction after mastectomy for cancer were not included in this study, so the authors included only first breast cancers in the control group. Since women with implants placed primarily in the 1960s and 1970s formed the study population, the effectiveness of mammography must also be evaluated. Mammographic technique prior to 1980 was inferior to modern mammography.
Some authors have raised concerns that calcification in implant capsules, either with the implant in place or after explantation, might lead to false-positive diagnoses of malignancy and unnecessary additional diagnostic or even therapeutic interventions. Worse, false-negative assumptions that calcifications are capsular instead of cancer associated might be made (see Chapter 5; see also Douglas et al., 1991; Fajardo et al., 1995; Gumucio et al., 1989; Leibman and Kruse, 1993; Peters and Smith, 1995; Peters et al., 1995d; Reynolds et al., 1994; Silverstein et al., 1990b, 1992; Stewart et al., 1992). Peters et al. (1998), Raso et al. (1999) and Rolland et
Page 284
al. (1989b) analyzed capsular calcifications and found them to be calcium phosphate, similar to cancerous calcifications. Rolland et al. (1989b) also reported zinc, which they speculated came from the implants, but Peters et al. (1998) and Raso et al. (1999) did not find this and pointed out that surrounding tissue probably contains much more zinc as a source for possible zinc-containing deposits than a breast implant. Consideration of whether calcifications are low or high density and comparison (in ex-planted breasts) with preexplant mammograms, among other things, may help to differentiate capsular from cancer-associated calcification (Raso et al., 1999; Rolland et al., 1989b). The above suggests that capsular calcification would constitute a strong indication for capsulectomy at the time of explantation (Peters et al., 1995d).
Although silicone breast implants definitely obscure some of the breast tissue which in theory at least might reduce visualization of breast tumors, the committee noted that some studies of cancer detection in women with silicone breast implants found the implants hindering and some studies found them not hindering detection. Special attention to detection is required in women with silicone breast implants. No studies have addressed whether there is differential mortality from breast cancer due to any differences in detection in women with or without breast implants.
Conclusions
The committee finds magnetic resonance imaging to be the most accurate imaging modality for the detection of intra- and extracapsular rupture. Mammography is of limited usefulness in detecting implant rupture in women with silicone implants. There is scant anecdotal evidence of rupture during mammography, and there are no data to support limiting screening or diagnostic mammography which would otherwise be indicated because of this concern. Implants placed in a subpectoral position do not interfere with mammography to the same extent as subglandular implants. Data on whether cancer detection is impaired by implants do not allow definite conclusions, although it is clear that implants do interfere with screening mammography by obscuring a variable part of breast tissue, distorting breast architecture, and especially in the presence of firm contractures, making a proper examination with proper compression of the breast more difficult and occasionally impossible.