Nuclear Magnetic Resonance and Detection of Human Breast Tumors 1 W. M. M. J. Bovee, 2 K. W. Getreuer,2 J. Smldt,2 and J. Lindeman ABSTRACT -Malignant, benign, and normal breast tissues corrected for fat appeared to have similar nuclear magnetic proton splnlattice relaxation times, contrary to previously published reports. However, due to a lower fat content, benign and malignant tumors relaxed more slowly than the surrounding normal tissue. This provides a possibility for locating them (but not for discriminating between them) In a breast by the nuclear magnetic resonance Imaging technique zeugmatography.-J Natl Cancer Inst 61: 53-55, 1978.
TABLE
I.-Longitudinal relaxation time Til of human mammary tissues at 60 MHz and 24° C. arranged in order of increasing magnitude"
Patient Til, msec No.
Patient No.
T
II,
msec
Patient No.
Til, msec
54
52 56 55 56 56 57 56 52 56 56
-
970 970 970 980 1,000 1,000 1,010 1,065 1,090 1.130 1.160 -
65 60 50 50 60 61 64 65 61 61 50 61 65 51 -
940 940 950 960 960 970 980 980 1,010 1,015 1,020 1,030 1,170 1,300 -
Malignant ti88ues
670 690 740 760 760 790 800 820 840 850 850 860
52 52 51 58 52 58 54 55 54 55 55 55
870 870 900
910 910 950 950 950 950 950 955 960
Nonmalignant tissues
65 59 59 50 59 55 55 50 50 50 55 50 59 64 62
MATERIALS AND METHODS
The TI values were determined as described elsewhere (4), at 24° C on a Bruker CXP4-100 spectrometer at 60 MHz. Human breast tissues were obtained from surgical pathology laboratories within an hour after surgery and stored up to 12 hours at 4° C. About 0.1-0.2 cm 3 of tissue was dissected and packed in 3.S-mm glass tubes. The selection of the samples was done by gross examination; the classification was based on the pathologic examination of the corresponding clinical case. The samples were classified as malignant or nonmalignant. During the investigation it appeared necessary to consider more carefully specific tissue samples on which the NMR measurements were made. Therefore, 4 of every 10 pieces were examined histologically after the NMR measurements had been made.
630 650 650 720 730 760 760 760 770 780 800 800 810 810 820
65 50 57 50 65 59 62 51 51 50 50 63 62 61 61
830 840 850 850 860 860 870 870 880 890 900 900 920 930 935
a Values of TIs have not been included because they all are ...zOO msec. More than one sample was usually taken from each patient.
fat. In agreement with this it appeared that the relaxation was not single valued. We discerned 2 fractions, fat and other tissue, with relaxation times TIs ABBREVIATIONS USED: NMR=nuclear magnetic resonance; T I • TIs. Til. and Tla=proton magnetic relaxation times. Received September 19. 1977; accepted November 25. 1977. Department of Applied Physics. Delft University of Technology. Post Box 5046. 2600 GA/Delft. The Netherlands. 5 Laboratory of Pathology SSDZ. R. de GraHweg 7. Delft. The Netherlands. • We thank the Laboratory of Pathology SSDZ; Drs. C. A. Broodman and M. A. Verschuyl. Bethel Hospital. Delft; Drs. J. W. Eckhardt and F. A. Tjebbes, Oude en Nieuwe Gasthuis, Delft; Dr. H. P. J. M. Platenburg, St. Clara Hospital. Rotterdam; Drs. J. c. J. Wereldsma and J. A. J. Spaas, Zuiderziekenhuis. Rotterdam; Dr. H. S. Posthuma. Holy Hospital. Vlaardingen; and Dr. H. S. Reinhold. Radiobiological Institute TNO, Rijswijk. for supplying the samples and for their interest. I
2
RESULTS
Several samples that appeared to be normal on gross examination contained small areas with benign dysplastic changes. Only nonmalignant samples containing none of these changes were called normal. In the adult female breast, connective tissue (ligamenta suspensoria mammae) and glandular tissue with supporting connective tissue are normally heavily admixed with
53
J
NATL CANCER INST
Downloaded from http://jnci.oxfordjournals.org/ at City University, London on April 16, 2016
53 53 53 58 58 50 52 58 58 54 54 58
Damadian (1) was the first to suggest NMR as a diagnostic tool for the detection of cancer, because the TI value is larger in cancerous tissues than in normal ones. This result has been confirmed by many other investigators (2). In 1973 Lauterbur (3) introduced a new NMR technique, zeugmatography, which makes possible the determination of spatial distribution of TI values in a given object, and hence (in view of Damadian's findings) the location of a malignant tumor in a body. The great need for a fast, noninvasive, and harmless method of detection and the high incidence of breast cancer prompted us to investigate the relaxation behavior and spatial distribution of TI values in normal, benign, and malignant human breast tissues. Moreover" zeugmatography is easier to perform on a part of, than on the whole, human body.
VOL. 61. NO. I. JULY 1978
3,4
54 Bovee, Getreuer, Smidt, et al.
5
t
1Il
+-'
2
C ::::J
>-
.'a
'-
+-'
.Q
'-
a
CII i:l ::::J
.....
.5
a.
E
a
01
iii
.2
time (miLliseconds) _ TEXT·HGl1RE I.-Relaxation curve of normal breast tissue of a sample taken from patient #62. The signal amplitude was proportional to the deviation from equilibrium of the proton magnetization in the sample. Til was determined from that part of the curve (larger values of the time) where only still signals of the slowly relaxing fatless tissue were present. By extrapolating these signals to smaller values of the time and subtracting them from the experimental points (e). the points (0) determining the relaxation of the fat were found. The fraction of fat was about one·third; ms=millisecond.
(with a small value of about 200 msec) and Til (with a larger value of about 900 msec), respectively. (The relaxation of the fat itself was only to a high degree of approximation single valued.) The signals of the fat and other tissue (including cysts) could also be seen in the NMR spectrum, where they are 220 Hz apart at 60 MHz. From the lOtal magnetization recovery curve (after a perturbation of the magnetization), Tis, Til and the fractions of fat (x) and other tissue (I-x) can be determined. An example is given in text-figure 1. The time constant Tla in this figure describes the first part of the magnetization-recovery curve and is given by: Tl~ = xTl~+(I·x)Tl\.
The value of Tla strongly depends on the fat content x. The Til values are given in tables I and 2 and text-figure
2.
TABLE 2.-Longitudinal relaxation time Til of human mammary
tissues at 24° C·
Kind of tissue Malignl>nt Nonmalignant Nonmalignant if 2 deviating samples are excluded Normal
Til, msec b
P, %C
911±20 (35) 879±19 (44) 862±16 (42)
25
TABLE
I.-Longitudinal relaxation time Til of human mammary tissues at 60 MHz and 24° C. arranged in order of increasing magnitude"
Patient Til, msec No.
Patient No.
T
II,
msec
Patient No.
Til, msec
54
52 56 55 56 56 57 56 52 56 56
-
970 970 970 980 1,000 1,000 1,010 1,065 1,090 1.130 1.160 -
65 60 50 50 60 61 64 65 61 61 50 61 65 51 -
940 940 950 960 960 970 980 980 1,010 1,015 1,020 1,030 1,170 1,300 -
Malignant ti88ues
670 690 740 760 760 790 800 820 840 850 850 860
52 52 51 58 52 58 54 55 54 55 55 55
870 870 900
910 910 950 950 950 950 950 955 960
Nonmalignant tissues
65 59 59 50 59 55 55 50 50 50 55 50 59 64 62
MATERIALS AND METHODS
The TI values were determined as described elsewhere (4), at 24° C on a Bruker CXP4-100 spectrometer at 60 MHz. Human breast tissues were obtained from surgical pathology laboratories within an hour after surgery and stored up to 12 hours at 4° C. About 0.1-0.2 cm 3 of tissue was dissected and packed in 3.S-mm glass tubes. The selection of the samples was done by gross examination; the classification was based on the pathologic examination of the corresponding clinical case. The samples were classified as malignant or nonmalignant. During the investigation it appeared necessary to consider more carefully specific tissue samples on which the NMR measurements were made. Therefore, 4 of every 10 pieces were examined histologically after the NMR measurements had been made.
630 650 650 720 730 760 760 760 770 780 800 800 810 810 820
65 50 57 50 65 59 62 51 51 50 50 63 62 61 61
830 840 850 850 860 860 870 870 880 890 900 900 920 930 935
a Values of TIs have not been included because they all are ...zOO msec. More than one sample was usually taken from each patient.
fat. In agreement with this it appeared that the relaxation was not single valued. We discerned 2 fractions, fat and other tissue, with relaxation times TIs ABBREVIATIONS USED: NMR=nuclear magnetic resonance; T I • TIs. Til. and Tla=proton magnetic relaxation times. Received September 19. 1977; accepted November 25. 1977. Department of Applied Physics. Delft University of Technology. Post Box 5046. 2600 GA/Delft. The Netherlands. 5 Laboratory of Pathology SSDZ. R. de GraHweg 7. Delft. The Netherlands. • We thank the Laboratory of Pathology SSDZ; Drs. C. A. Broodman and M. A. Verschuyl. Bethel Hospital. Delft; Drs. J. W. Eckhardt and F. A. Tjebbes, Oude en Nieuwe Gasthuis, Delft; Dr. H. P. J. M. Platenburg, St. Clara Hospital. Rotterdam; Drs. J. c. J. Wereldsma and J. A. J. Spaas, Zuiderziekenhuis. Rotterdam; Dr. H. S. Posthuma. Holy Hospital. Vlaardingen; and Dr. H. S. Reinhold. Radiobiological Institute TNO, Rijswijk. for supplying the samples and for their interest. I
2
RESULTS
Several samples that appeared to be normal on gross examination contained small areas with benign dysplastic changes. Only nonmalignant samples containing none of these changes were called normal. In the adult female breast, connective tissue (ligamenta suspensoria mammae) and glandular tissue with supporting connective tissue are normally heavily admixed with
53
J
NATL CANCER INST
Downloaded from http://jnci.oxfordjournals.org/ at City University, London on April 16, 2016
53 53 53 58 58 50 52 58 58 54 54 58
Damadian (1) was the first to suggest NMR as a diagnostic tool for the detection of cancer, because the TI value is larger in cancerous tissues than in normal ones. This result has been confirmed by many other investigators (2). In 1973 Lauterbur (3) introduced a new NMR technique, zeugmatography, which makes possible the determination of spatial distribution of TI values in a given object, and hence (in view of Damadian's findings) the location of a malignant tumor in a body. The great need for a fast, noninvasive, and harmless method of detection and the high incidence of breast cancer prompted us to investigate the relaxation behavior and spatial distribution of TI values in normal, benign, and malignant human breast tissues. Moreover" zeugmatography is easier to perform on a part of, than on the whole, human body.
VOL. 61. NO. I. JULY 1978
3,4
54 Bovee, Getreuer, Smidt, et al.
5
t
1Il
+-'
2
C ::::J
>-
.'a
'-
+-'
.Q
'-
a
CII i:l ::::J
.....
.5
a.
E
a
01
iii
.2
time (miLliseconds) _ TEXT·HGl1RE I.-Relaxation curve of normal breast tissue of a sample taken from patient #62. The signal amplitude was proportional to the deviation from equilibrium of the proton magnetization in the sample. Til was determined from that part of the curve (larger values of the time) where only still signals of the slowly relaxing fatless tissue were present. By extrapolating these signals to smaller values of the time and subtracting them from the experimental points (e). the points (0) determining the relaxation of the fat were found. The fraction of fat was about one·third; ms=millisecond.
(with a small value of about 200 msec) and Til (with a larger value of about 900 msec), respectively. (The relaxation of the fat itself was only to a high degree of approximation single valued.) The signals of the fat and other tissue (including cysts) could also be seen in the NMR spectrum, where they are 220 Hz apart at 60 MHz. From the lOtal magnetization recovery curve (after a perturbation of the magnetization), Tis, Til and the fractions of fat (x) and other tissue (I-x) can be determined. An example is given in text-figure 1. The time constant Tla in this figure describes the first part of the magnetization-recovery curve and is given by: Tl~ = xTl~+(I·x)Tl\.
The value of Tla strongly depends on the fat content x. The Til values are given in tables I and 2 and text-figure
2.
TABLE 2.-Longitudinal relaxation time Til of human mammary
tissues at 24° C·
Kind of tissue Malignl>nt Nonmalignant Nonmalignant if 2 deviating samples are excluded Normal
Til, msec b
P, %C
911±20 (35) 879±19 (44) 862±16 (42)
25
