乳腺密度（二）
测定乳房密度
几十年来，放射科医生对乳房密度进行了主观分级。我认为我们在识别腺体分型方面相当准确，但声称这些分型可以转化为实际的腺体体积测量是无稽之谈。我在2008年的文章¹中对此进行了解释。
现在的确可以实际测量数字乳腺X射线摄影检查成像视野中纤维腺体组织的数量。通过了解乳房的厚度、mAs和kVp，以及透过乳房到达特定探测器元件的辐射量，可以计算出该探测器元件上方的“组织柱”中纤维和脂肪组织的百分比。通过将所有的柱状测量值加在一起，系统可以计算出“致密”（水）组织的体积和脂肪的数量。然而，这只能告诉你致密组织的体积占图像视野内组织总体积的百分比。由于我们没有对整个乳房进行成像，也由于我们实际上不知道乳房的边界在哪里，所以不可能对乳房的实际密度百分比进行任何精确地测量。
此外，为什么一个小乳房、致密度70%、纤维腺组织体积为10cc（ACR分类“d”）的女性比一个乳房大得多、拥有散布在整个乳房体积远远超过10cc的纤维腺体、ACR分类“b”的女性面临更高的乳腺癌风险？
乳房密度和乳腺癌的风险
我不认为组织形态是衡量患乳腺癌风险的有效方法。之所以有数据提示如此，可能是由于在那些筛查研究中，如果女性做了乳腺X射线摄影检查，那些具有脂肪型乳房并被诊断为乳腺癌的女性，很可能在筛查项目早期或刚开始进行筛查的时候就被检测到了，而在筛查项目后期发现的癌症很可能是隐藏在致密乳腺组织中的癌症。
事实上，将乳腺密度与乳腺癌风险联系起来的研究非常可疑。他们都试图测量乳房中致密组织的百分比与脂肪的含量，然后将其与乳腺癌风险联系起来。如上所述，所有这些估计都是因为乳房没有可定义的边界，乳腺的终点在哪里？如果不能准确测量乳房的总体积，就不可能测量致密体积的百分比（体积百分比=致密组织体积/乳房总体积）。
关于致密组织导致风险增加的说法被大大夸大了。首先，绝大多数（80%）40岁以下的女性拥有致密的乳腺组织，但这些女性患乳腺癌的风险最低。乳房致密的女性比例随着年龄的增长而下降²。绝经后服用补充激素的女性变得“脂肪替代”的速度较慢。有一些数据表明，绝经后使用激素会略微增加患乳腺癌的风险。也许致密乳腺组织的持续存在是偶然的。
在我们的数据中，有特定分型的女性在人口中所占的比例与有每种分型的乳腺癌诊断女性所占的比例相似。问题在于，致密的组织可以隐藏癌症，而癌症检测的延迟很可能是风险增加的原因（癌症在脂肪乳中更早发现）。同样，当大多数20多岁和30岁出头的女性拥有“致密”乳房，而她们患乳腺癌的风险最低时，认为致密乳房的风险更高，这是自相矛盾的。也许，出现致密乳腺组织能够持续到绝经后多年这一现象，本身就是风险（激素的原因），但这目前还不能确定。
在我们麻省总院的数据中，乳腺组织形态分布与ACR的分布数字有些不同，但我们的数据来自于屏片时代。
a. 9%的妇女“a”型
b. 24%的妇女“b”型
c. 56%的妇女“c”型
d. 11%的妇女“d”型
根据我们的经验，癌症的比例与上述分布大致相似。女性乳腺组织形态的分布：
a. “a”型占所有女性的9%，占被诊断为癌症的女性的4%
b. “b”型占所有女性的24%，占被诊断为癌症的女性的23%
c. “c”型占所有女性的56%，占被诊断为癌症的女性的64%
d. “d”型占所有女性的11%，占被诊断为癌症的女性的9%
ACR乳腺类型为“c”和“d”的女性约占总人口的67%，占乳腺癌的72%。然而，乳腺分类为“a”或“b”的女性（约占总人口的33%），如果仅仅因其风险稍低而被排除在筛查之外，那么我们将遗漏掉约27%的癌症。
最近使用数字乳腺X射线摄影的麻省总院的数据³结果类似：
a. 脂肪型乳房（ACR“a”）的女性占人口的7.6%，占癌症的5.7%
b. 散在腺体分布型乳房（ACR“b”）的女性占人口的47.7%，占癌症的47.2%
c. 不均匀分布腺体型乳房（ACR“c”）的女性占人口的38.9%，占癌症的42.2%
d. 极高腺体密度型乳房（ACR“d”）的女性占人口的4.9%，占癌症的4.9%
根据我们的经验，“极高腺体密度（“d”）”乳房的女性，其乳腺癌风险仅略高于“完全脂肪型”乳房的女性，甚至根本没有升高。而大多数被诊断为乳腺癌的女性具有“b”型和“c”型。根据我们的经验（其他人也有类似的经验），致密乳腺组织不会特别增加患乳腺癌的风险，但毫无疑问，致密组织可以在乳腺X射线摄影图像上遮挡住癌症。这些癌症有些在超声影像上可见，更多的在MRI影像上可见。但是在我们能够真正有效使用超声和MRI筛查乳腺癌之前，数字乳腺X射线摄影仍然是检测早期乳腺癌的主要手段。
总结
乳腺密度不是主要的危险因素，但正常的乳腺纤维腺组织可能会遮挡癌症。数字乳腺体层合成（DBT）的使用能够去除重叠的组织，从而将癌症的检出率提高约30%，但如果癌症实际上生长在纤维腺组织中，它就可能被遮挡。这种情况下的癌症可以通过超声或MRI更好地检测出来。
英文对照｜ENGLISH COMPARISON
BREAST DENSITY-LOWER
DETERMINING BREAST DENSITY
For decades breast density was subjectively graded by radiologists. I think we were fairly accurate at recognizing patterns, but claims that these could be translated into actual volumetric measures were nonsense. This is explained in my article in 2008¹.
It is now possible to actually measure the amount of fibroglandular tissue in the field of view on a digital mammogram. By knowing the thickness of the breast, the mAs, and kVp, and the amount of radiation passing through the breast reaching a given detector element, the percentages of fibrous and fatty tissue, in the column of tissue overlying that detector element, can be calculated. By adding all the column measures together, the system can calculate the volume of tissue that is “dense” (water) and the amount that is fat. However, this only tells you the volume of dense tissues as a percent of the total volume of tissues in the field of view of the image. Since we do not image the entire breast, and since we actually have no idea where the breast ends, any accurate measure of the actual percent density of the breast is not possible.
Furthermore, why would a woman with a small breast that is 70% dense with a volume of 10 cc of fibroglandular tissue (ACR pattern “d”), be at higher risk than a woman with a much larger breast that has tissues scattered throughout the fat making it an ACR pattern “b”, but containing far more than 10 cc’s of fibroglandular tissue?
DENSITY AND RISK OF DEVELOPING BREAST CANCER
I do not think that tissue patterns are a useful measure of risk for developing breast cancer. The data that suggest this are probably due to the fact that in any study, if the women have had mammograms, those who have fatty breasts and have been diagnosed with breast cancer, have likely been detected early in the program, or prior to the start of the program. Consequently, the cancers that are found during (later in) the program will, more than likely, be those that were hidden in dense breast tissue.
In fact, studies linking density to risk are highly suspect. They have all tried to gauge the percent of the breast that is dense tissue compared to the amount of fat, and then associate that with breast cancer risk. As noted above, all such estimates suffer from the fact that the breast has no definable boundaries–where does it end? Without being able to accurately measure the total volume of the breast, it is impossible to measure the percent of the volume that is dense (percent volume=volume of dense tissue/total volume of the breast).
Claims of increased risk due to dense tissue are greatly exaggerated. In the first place, the vast majority (80%) of women under the age of 40 have dense breast tissues, yet these women have the lowest risk of developing breast cancer. The percent of women with dense breasts decreases as women age². Women who are using supplemental hormones after menopause are slower to become “fatty replaced”. There are some data that suggest that using hormones after menopause slightly increases the risk of breast cancer. Perhaps the persistence of dense breast tissue is coincident.
In our data, the same percentage of women in the population that have a particular pattern is similar to the percent of women diagnosed with breast cancer who have each pattern. The problem is that dense tissues can hide cancers and the delay in their detection is likely the source of claims of increased risk (cancers are found earlier in fatty breasts). Again, it is paradoxical to suggest that dense breasts are at higher risk when most women in their twenties and early thirties have “dense” breasts yet they have the lowest risk of breast cancer. Perhaps it is the persistence of dense breast tissue into the postmenopausal years that is a risk (hormones), but this is not known.
At the Massachusetts General Hospital our distribution of tissue patterns was somewhat different from the ACR numbers but our data were from the era of screen/film:
a. 9% of all women had pattern “a”.
b. 24% of all women had pattern “b”
c. 56% of all women had pattern “c”
d. 11% of all women had pattern “d”
In our experience the percentage of cancers was similar to the above distribution. Women with pattern:
a. Accounted for 9% of all women and 4% of the women diagnosed with cancer
b. Accounted for 24% of all women and 23% of the women diagnosed with cancer
c. Accounted for 56% of all women and 64% of the women diagnosed with cancer
d. Accounted for 11% of all women and 9% of the women diagnosed with cancer
Women with ACR patterns “c” and “d” accounted for approximately 67% of the population and 72% of the cancers. However, if women with ACR PPAT “a” or “b” (approx.33% of the population)had been excluded from screening because they are at slightly lower risk, we would overlook approximately27% of the women with cancer.
More recent MGH data³using Digital Mammography are similar:
a. Women with fatty breasts (ACR “a”) make up 7.6% of the population and account for 5.7% of the cancers
b. Women with scattered density (ACR “b”) make up 47.7% of the population and 47.2% of the cancers
c. Women with heterogeneous tissues (ACR “c”) make up 38.9% of the population and 42.2% of the cancers
d. Women with extremely dense tissues (ACR “d”) make up 4.9% of the population and 4.9% of the cancers
In our experience women with “extremely dense tissues (“d”)” were only at slightly elevated risk to not elevated at all, when compared to women with “all fat”, and the most women diagnosed with cancer had patterns “b” and “c”. Based on our experience (others have had similar experience), dense tissues are not a particularly increased risk of developing breast cancer, but there is no question that dense tissues can hide cancers on mammograms. Some are visible on ultrasound, and even more are visible on MRI. Until we can screen effectively and efficiently using ultrasound and MRI, mammogrpahy remains the mainstay of early detection.
SUMMARY
Breast density is not a major risk factor, but normal fibroglandular breast tissues can hide cancers. The use of Digital Breast Tomosynthesis (DBT) increases the cancer detection rate by about 30% by being able to remove superimposed tissues, but if a cancer actually grows in fibroglandular tissues it can be hidden. These cancers can be better detected using Ultrasound or MRI.
REFERENCES
1.Kopans DB.Basic physics and doubts about relationship between mammographically determined tissue density and breast cancer risk. Radiology. 2008;246:348-53.
2.Stomper PC, D'Souza DJ, DiNitto PA, Arredondo MA. Analysis of Parenchymal Density on Mammograms in 1353 Women 25-79 Years Old. AJR 1996;167:1261-1265.
3.Yala A, Lehman C, Schuster T, Portnoi T, Barzilay R. A Deep Learning Mammography-based Model for Improved Breast Cancer Risk Prediction. Radiology. 2019 Jul;292(1):60-66.
