The British Journal of Radiology, 82 (2009), 640–644

Significance of tumour calcification in ovarian carcinoma 1

G J C BURKILL, BSc, MRCP, D M KING, DMRD, FRCR

FRCR,

1

S D ALLEN,

MRCS, FRCR,

2

R P A’HERN,

MSc,

3

M E GORE,

PhD, FRCP

and

1

Departments of 1Radiology, 2Medical Statistics and 3Medical Oncology, The Royal Marsden Hospital, Fulham Road, London SW3 6JJ, UK

ABSTRACT. The purpose of this study was to assess the pattern and significance of tumour calcification in ovarian carcinoma. Patients with calcifying ovarian carcinoma were identified from radiological reports. Their tumour characteristics, serum calcium levels, treatment and survival were compared with a control group of patients with non-calcifying disease. Patterns and distribution of calcification were assessed. Available serial CT scans were reviewed for changes in both soft-tissue and calcified disease according to RECIST (response evaluation criteria in solid tumours) criteria where feasible. Temporal changes in calcification were correlated with changes in soft tissue disease and CA125 levels. The calcified group numbered 122 (22 other patients had calcifying tumour but insufficient clinical data). Calcification in ovarian carcinoma had a prevalence of 8% (144/1721) in our series. There was a significant difference (p,0.001) between the two groups in the distribution of histological type, with serous tumours being more common in the calcified group (74/122 (61%)) than in the controls (509/1498 (34%)). The calcified tumour patients tended to have lower grade disease (p,0.001). No differences between the groups were found for age, treatment or serum calcium levels. Distribution of calcification was diffusely peritoneal in 34 patients, in association with a pelvic mass in 15, nodal in 11 and within the anterior abdominal wall in 2. There was no correlation between changes in calcification on serial CT scans and corresponding CA125 levels. In conclusion, calcification tends to occur most commonly in serous cystadenocarcinomata and in tumours of lower grade. Changes in calcification cannot be used as a marker of disease response.

Calcification has been reported in a wide range of both primary neoplasms and metastatic diseases. The interest in calcification in the radiological literature has largely centred on its identification in many disparate conditions with warnings of its unreliability as a predictor of benign disease [1–3]. A number of studies have attempted to define the significance of malignant tumour calcification. In lung carcinoma, calcification occurs more commonly when the tumour is large and centrally placed in the lung, and is independent of histological subtype [4]. In neuroblastoma, calcification is common but its presence does not influence prognosis [5]. Two reports on the prognostic influence of calcification in colorectal liver metastases reached different conclusions. In one, calcification predicted longer survival, which was independent of other variables [6]. The second report found that calcification conferred a small negative impact on survival that was not statistically significant [7]. Calcification in renal tumours is independent of tumour grade and once again is unlikely to affect prognosis [3]. Calcification in ovaries is well described and can be attributed to non-neoplastic conditions such as infarction and mature teratoma; it can also be an incidental finding in an otherwise normal ovary and may be the only indication of neoplasia [8–10]. Calcification in ovarian Address correspondence to: S D Allen, Department of Radiology, The Royal Marsden Hospital, Downs Road, Sutton, Surrey SM2 5PT, UK. E-mail: [email protected]

640

Received 8 April 2008 Revised 24 June 2008 Accepted 14 July 2008 DOI: 10.1259/bjr/12716831 ’ 2009 The British Institute of Radiology

carcinoma metastases has long been recognised radiographically, particularly in serous adenocarcinoma [11, 12]. It has been described in both intra- and extraabdominal deposits, including brain metastases [13, 14]. Its presence on the peritoneal surfaces, the most common manifestation, can aid the radiologist in detecting small deposits [12]. To our knowledge, however, no previous studies have attempted to determine the significance of tumour calcification in ovarian carcinoma. We reviewed cases of calcifying ovarian carcinoma at our institution, comparing them with a non-calcifying set of controls. The aim of this study was to assess the pattern and significance of tumour calcification in ovarian carcinoma.

Methods and materials This study commenced at a time when ethics committee approval was not required for retrospective analyses of clinically available data, therefore approval was not sought. This study was conducted in accordance with the Helsinki Declaration of 1975, as revised in 1983. The radiological reports of all patients at our institution with a histological diagnosis of ovarian carcinoma were searched for words commencing with the abbreviation ‘calcif’. The full reports of all examinations that included such words were reviewed to establish whether the calcification related to the patient’s tumour on CT The British Journal of Radiology, August 2009

Tumour calcification in ovarian carcinoma

scanning. All patients with tumour calcification and available clinical data were analysed. A ‘‘non-calcified’’ group was drawn from those patients with ovarian carcinoma who had had CT scans but no calcification reported. The calcified and non-calcified groups were compared for histological type, stage at diagnosis, tumour grade, serum calcium level at presentation, type of chemotherapy, the use of radiotherapy and survival. Two radiologists (G.J.C.B. and D.M.K.) reviewed the CT images of the calcified group by consensus. The following information was recorded: the date of the first appearance of calcification, the patterns of calcification and changes in the extent of calcification, and soft-tissue disease over sequential scans. Calcification pattern was categorised into: punctate well-circumscribed homogeneous nodules, a linear or curvilinear thin straight pattern (curved lines or sheets), and amorphousshapeless heterogeneous conglomerations. The response evaluation criteria in solid tumours (RECIST) criteria were applied independently for changes in disease status, both for soft tissue deposits and for calcification. When this was not possible, a judgement of best response was made and the disease was categorised as complete response, partial response, stable disease or progressive disease [15]. These data were then compared to changes in cancer antigen 125 (CA125) level, which were classified as a response according to the method described by Rustin et al. [16]. CT scans and CA125 levels were assumed to be coincident if they were performed within 28 days of each other. Statistical analysis was performed using the x2 and Mann–Whitney U tests. A prognostic factor analysis and multivariate analysis (Cox regression) were also performed on the data in order to examine the effect of histological subtype, tumour stage and tumour grade.

Results A total of 144 patients with confirmed carcinoma of the ovary had calcifying lesions on CT scanning, of which 122 had follow-up data on the Gynaecological Unit’s database maintained for research purposes. These 122 cases made up the calcified group. The non-calcified group comprised 1577 patients who had the same diagnosis, whose data was on the same database and

Table 1. Distribution of patients according to histological types Histology

Non-adenocarcinoma Mucinous Serous Endometrioid Mesonephroid Cytadenocarcinoma (unspecified) Adenocarcinoma (unspecified) Mixed Total a

Non-calcified

Calcified

119 (8%) 165 (11%) 509 (34%) 165 (11%) 60 (4%) 105 (7%)

5 (4%) 7 (6%) 74 (60%) 2 (2%) 2 (2%) 5 (4%)

360 (24%)

22 (18%)

15 (1%) 1498a

5 (4%) 122

Data on histological subtype were not available for 79 of the original 1577 non-calcified patients, leaving a total of 1498 patients in this group.

who were followed up contemporaneously with the calcified group. Therefore, tumour calcification had a prevalence of (144/1721) 8% in this study. The mean ages at presentation for the calcified and non-calcified groups were 55 years (range 9–88 years) and 53 years (range 21– 84 years), respectively. The distribution of patients according to the year of diagnosis was similar for both groups. There were significant differences in the distribution of histological type between the two groups with a greater predominance of serous tumours in the calcified population (p,0.001) (Table 1). The calcified group presented with higher stage disease, with (111/ 122) (91%) being stage III or IV, whereas 853/1421 (60%) were of equivalent stage in the non-calcified group (p,0.001) (Table 2). Conversely, the calcified group tended to have tumours of significantly lower grade (p,0.001) (see Table 3). The calcified group had a poorer survival rate, with median survival 61 months compared to 132 months for the non-calcified group (hazard ratio 1.49 (95% confidence interval (CI): 1.10–2.01), p50.002). CA125 response had no significant relationship with calcification regression (p50.24). There was no difference in the mean serum calcium levels at presentation of the calcified (2.36 mmol l–l (SD 0.02)) and non-calcified (2.39 mmol l–l (SD 0.15)) groups, p50.49. Elevated serum calcium was recorded at presentation in 5.6% of non-calcified patients and 1.3% of calcified patients (p50.19).

Figure 1. Examples of calcification patterns in metastatic ovarian carcinoma. (a) Amorphous calcification over the surface of the liver (arrow) and spleen. (b) Punctate (arrows) and (c) linear calcifications (arrows) on the underside of the diaphragm. The British Journal of Radiology, August 2009

641

G J C Burkill, S D Allen, R P A’Hern et al Table 2. Distribution of patients according to stage at presentation Stage

Non-calcified

Calcified

I II III IV Total

426 (30%) 142 (10%) 668 (47%) 185 (13%) 1421a

7 (6%) 4 (3%) 89 (73%) 22 (18%) 122

a

Figure 2. Extensive calcification associated with a large pelvic recurrence. The architecture of the calcification is better appreciated on bone window settings.

There were only minor differences in the chemotherapy regimens of non-calcified and calcified patients, with a greater percentage of the non-calcified group receiving chemotherapy compared with the calcified patients. There was no difference in the use of radiotherapy between the two groups. The multivariate analysis showed that only cancer stage and grade were significant predictors of outcome (both p,0.001), implying that the poor prognosis observed in patients with calcification was attributable to its association with stage. Imaging examinations were available for review for 48 of the 122 calcified group patients. There was a median of 6 CT scans per patient (range 2–17). Three patterns of calcification were identified, namely amorphous, punctate and linear/curvilinear (Figure 1). The majority of patients 36/48 (75%) exhibited a single form of calcification: amorphous in 20, punctate in 12 and linear in 4. The remaining 12 patients demonstrated mixed patterns of calcification. The distribution of calcification was most commonly diffusely peritoneal (n534). The other sites were as follows: in association with a pelvic mass (n515)

Figure 3. Amorphous calcification within an enlarged retroperitoneal lymph node (arrow).

642

Staging data not available for 156 patients of the original 1577 non-calcified patients, leaving a total of 1421 patients in this group.

(Figure 2), within enlarged lymph nodes (n511) (Figure 3) and within the anterior abdominal wall (n52). Calcification was present on the first available CT scan in 33 of the 48 patients. In 15 of these, the first available scan pre-dated first-line chemotherapy. Calcification developed after the initiation of chemotherapy in 10 patients. For the remaining eight patients, the temporal relationship of imaging and treatment could not be established as the patients had been referred to our institution for further management and their prior imaging was unavailable. In three patients, calcification regressed during followup in the absence of surgical intervention. In two cases, the regression coincided with a soft-tissue disease response during chemotherapy (Figure 4), whereas in the third, it was incidental to progression of soft-tissue disease approximately 1 year after completion of firstline chemotherapy.

Discussion There are a number of potential causes for the development of calcification in neoplastic disease. First, certain tumours have specific histological features that result in calcium deposition, the best known of these being the bone- and cartilage-derived sarcomas [17]. Pertinent to the ovary, but also seen in other carcinomas, is a papillary tumour morphology. These tumours form psammoma bodies, which are concentrically laminated calcific concretions [11, 17, 18]. Psammomatous calcification is likely to account for the majority of calcified disease in our study group as there was a predominance of serous cystadenocarcinomas, which often exhibit papillary projections histologically. Psammoma calcification is typically punctate. However, amorphous calcification, which was the most common pattern in our patients, is known to occur in psammomatous calcification where the psammoma bodies act as niduses for the deposition of new layers of calcification [17]. Calcification is not an uncommon finding in colorectal liver metastases, occurring in 12–28% of cases [6, 7]. Calcification in the metastases of this tumour has attracted much attention, but its remain poorly understood. It may simply be the result of tumour haemorrhage and necrosis. The proposed mechanism for calcification in cell death is that impaired cellular metabolism in necrotic tissue results in an elevation of pH, and thus a tendency for the precipitation of calcium salts [19]. Where amorphous calcifications resided in soft tissue masses in our patients, the necrosis/haemorrhage The British Journal of Radiology, August 2009

Tumour calcification in ovarian carcinoma

Figure 4. Serial CT scans demonstrating the regression of a calcified deposit (arrow) between the rectum and vagina during chemotherapy: (a) Pre-treatment baseline scan, (b) scan following 3 months of treatment and (c) scan following 6 months of treatment.

theory seems plausible, because the vascular supply to these relatively large lesions may become precarious. In biopsies obtained from calcified colorectal liver metastases, however, calcification is seen next to viable tumour cells without an adjacent inflammatory reaction [6]. Treatment, be it radiotherapy or chemotherapy, has also been implicated in tumour calcification. The pathological basis is presumably an accelerated version of the calcification seen in haemorrhage and necrosis mentioned above. Many cases of calcification following treatment have been reported; post-radiation therapy examples include lymphoma, ovarian metastases and prostate carcinoma [20–22]. Tumour calcification following chemotherapy has been reported in transitional cell carcinoma of the bladder following intravesical mitomycin C therapy and in regressing trophoblastic lung metastases following methotrexate and mercaptopurine [23, 24]. However, these mechanisms are unlikely to explain much of the calcification in our calcified group for a number of reasons. First, radiotherapy is infrequently used in carcinoma of the ovary. In terms of chemotherapeutic regimens, there were no differences between the treatments given to the calcified and noncalcified patients. Neither was there an inverse relationship between response to treatment, measured either on CA125 or in terms of soft-tissue disease, and the extent of calcification. Finally, a significant proportion of patients calcify before treatment. Another explanation of tumour calcification in colorectal carcinoma is mucinous degeneration, which is termed mucoid calcification. One study, however, failed to demonstrate a correlation between mucinous histology and calcification [6]. There was only a small number of mucinous tumours in our study, and the proportion of Table 3. Distribution of patients according to grade at presentation Grade

Non-calcified

Calcified

Borderline I II III Total

79 (7%) 91 (8%) 420 (37%) 544 (48%) 1134a

15 11 67 29 122

(12%) (9%) (55%) (24%)

a

Data on tumour grade were not available for 443 of the original 1577 non-calcified patients, leaving a total of 1134 patients in this group.

The British Journal of Radiology, August 2009

patients with this histology was greater in the noncalcified group. Owing to the controversy surrounding the above proposed mechanism for calcification in colorectal metastases, it has been proposed that the calcification is due to the secretion of an as-yet undefined tumour-specific factor [6, 7, 17]. Soft-tissue calcification is a feature of hyperparathyroidism. A wide variety of tumours are able to induce a metabolic state mimicking hyperparathyroidism through the release of a humoral factor. The most frequently implicated factor in humoral hypercalcemia of malignancy (HHM) is parathyroid hormone-related protein (PTHrP) [19, 25, 26]. Other responsible humoral factors that have been isolated include prostaglandins and 1,25dihydroxyvitamin D [27, 28]. This paraneoplastic syndrome is well described in ovarian carcinoma. Although reported in serous cystadenocarcinoma, endometrioid carcinoma and yolk sac tumours, it is most often found in association with clear cell adenocarcinoma and small cell carcinoma [25, 26, 29]. The absence of difference in serum calcium levels between the calcified and non-calcified groups, and the distribution of histological types in our calcified group, makes it unlikely that humoral hypercalcaemia of malignancy had a significant role in the development of calcification. In addition, HHM is usually a preterminal event and so is unlikely to result in significant soft-tissue calcification or to explain the presence of calcification early in the course of patients’ disease [26]. The differences in grade between the calcified and non-calcified groups can be explained by the fact that extensive papillarity is a finding in lower-grade serous carcinomas. Marked psammoma body formation is also a feature of well-differentiated tumours, as would be expected given that a certain level of organisation or maturity within the tumour is required to produce these foci of laminated calcification [30]. The difference in survival between the two groups was independent of calcification and could be accounted for by the differences in staging. We therefore need to account for the stage differences. The most likely explanation for this is group selection bias. In order for a patient to enter the calcified group, they had to have calcified metastatic disease as very few of patients at our institution (a tertiary referral centre for gynaecological malignancies) would have had imaging of the primary tumours by CT. By contrast, for a patient to enter the 643

G J C Burkill, S D Allen, R P A’Hern et al

non-calcified group, they simply needed a histological diagnosis of ovarian carcinoma and abdomino-pelvic imaging by CT. Metastatic disease was not a requirement for entry to the control group. A further potential problem with the study design is the requirement that calcification be mentioned in the radiology report in order for patients to be identified. This concern is tempered by the fact that all the radiologists at our institution were represented amongst the original radiological reports of the calcified group. Second, patients tended to have several scans with the calcification usually present relatively early in the course of the disease, giving ample opportunity for its description. At our institution, abdomino-pelvic CT scans for patients with ovarian carcinoma are acquired following oral water rather than positive contrast that might aid the detection of calcified deposits. In addition, as previously noted, it is not entirely appropriate to use RECIST criteria to evaluate calcified lesions [15]. As there is no other suitable or superior method to evaluate such lesions, however, RECIST criteria were used as the best available method to assessing disease response. This study shows the prevalence of calcification in a large series of ovarian tumours to be approximately 8%. Calcification occurs in many tumour types but predominantly in serous cystadenocarcinoma; hence psammoma bodies are the most likely source of tumour calcification. Tumour calcification in ovarian metastases is a dynamic process independent of treatment effects; therefore, changes in calcification cannot be used as a marker of disease status.

References 1. Mahoney MC, Shipley RT, Corcoran HL, Dickson BA. CT demonstration of calcification in carcinoma of the lung. AJR Am J Roentgenol 1990;154:255–8. 2. Eisenkraft BL, Som PM. The spectrum of benign and malignant aetiologies of cervical node calcification. AJR Am J Roentgenol 1999;172:1433–7. 3. Haddad FS, Shah IA, Manne RK, Costantino JM, Somsin AA. Renal cell carcinoma insulated in the renal capsule with calcification and ossification. Urol Int 1993;51:97–101. 4. Grewal RG, Austin JHM. CT demonstration of calcification in carcinoma of the lung. J Comput Assist Tomogr 1994;18:867–71. 5. Ross P. Calcification in liver metastases from neuroblastoma. Radiology 1965;85:1074–9. 6. Easson AM, Barron PT, Cripps C, Hill G, Guindi M, Michaud C. Calcification in colorectal hepatic metastases correlates with longer survival. J Surg Oncol 1996;63:221–5. 7. Hale HL, Husband JE, Gossios K, Norman AR, Cunningham D. CT of calcified liver metastases in colorectal carcinoma. Clin Radiol 1998;53:735–41. 8. Sheath S, Fishman EK, Buck JL, Hamper UM, Sanders RC. The variable sonographic appearances of ovarian teratomas: correlation with CT. AJR Am J Roentgenol 1988;151:331–4. 9. Fletcher RM, Boal DK, Karl SR, Gross GW. Ovarian torsion: an unusual case of bilateral pelvic calcification. Pediatr Radiol 1988;18:172–3. 10. Brandt KR, Thurmond AS, McCarthy JL. Focal calcifications in otherwise ultrasonographically normal ovaries. Radiology 1996;198:415–7.

644

11. Castro JR, Klein EW. The incidence and appearance of roentgenologically visible psammomatous calcification of papillary cystadenocarcinoma of the ovaries. AJR Am J Roentgenol 1962;88:886–91. 12. Mitchell DG, Hill MC, Hill S, Zaloudek C. Serous carcinoma of the ovary: CT identification of metastatic calcified implants. Radiology 1986;158:649–52. 13. Patel SV, Spencer JA, Wilkinson N, Perren TJ. Supradiaphragmatic manifestations of papillary serous adenocarcinoma of the ovary. Clin Radiol 1999;54:748–54. 14. Henriquez I, Castro C, Berenguer J, Biete A. Calcification of presumed ovarian carcinoma brain metastases following radiotherapy. Br J Radiol 1999;72:85–8. 15. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubinstein L, et al. New guidelines to evaluate the response to treatment in solid tumors. J Natl Cancer Inst 2000;92:205–16. 16. Rustin GJ, Nelstrop AE, McClean P, Brady MF, McGuire WP, Hoskins WJ, et al. Defining response of ovarian carcinoma to initial chemotherapy according to serum CA 125. J Clin Oncol 1996;14:1545–51. 17. Maile CW, Rodan BA, Godwin JD, Chen JTT, Ravin CE. Calcification in pulmonary metastases. Br J Radiol 1982;55:108–13. 18. Chen SW, Bennett G, Price J. Axillary lymph node calcification due to metastatic papillary carcinoma. Australas Radiol 1998;42:241–3. 19. Kunieda K, Okuhira M, Nakano T, Nakatani S, Tateiwa J, Hiramatsu A, et al. Diffuse calcification in gastric cancer. J Int Med Res 1990;18:506–14. 20. Fisher AMH, Kendall B, VanLeuvan BD. Hodgkin’s disease: a radiological survey. Clin Radiol 1962;13:113–27. 21. Menuck L. Intraabdominal calcification in treated disseminated carcinoma of the ovary. Obstet Gynecol 1977;49:56S–58S. 22. Jones WA, Miller EV, Sullivan LD, Chapman WH. Severe prostatic calcification after radiation therapy for cancer. J Urol 1979;121:828–30. 23. Drago PC, Badalament RA, Lucas J, Drago JR. Bladder wall calcification after intravesical mitomycin C treatment of superficial bladder cancer. J Urol 1989;142:1071–2. 24. Cockshott WP, Hendrickse JP. Pulmonary calcification at the site of trophoblastic metastases. Br J Radiol 1969;42:17–20. 25. Kitazawa R, Kitazawa S, Matui T, Maeda S. In situ, detection of parathyroid hormone-related protein in ovarian clear cell carcinoma. Hum Pathol 1997;28:379–82. 26. Tsunematsu R, Saito T, Iguchi H, Fukuda T, Tsukamoto N. Hypercalcemia due to parathyroid hormone-related protein produced by primary ovarian clear cell adenocarcinoma: case report. Gynaecol Oncol 2000;76:218–22. 27. Josse RG, Wilson DR, Heersche JNM, Mills JRF, Murray TM. Hypercalcemia with ovarian carcinoma. Evidence of a pathogenic role for prostaglandins. Cancer 1981;48:1233–41. 28. Hoekman K, Tjandra YI, Papapoulos SE. The role of 1,25dihydroxyvitamin D in the maintenance of hypercalcemia in a patient with an ovarian carcinoma producing parathyroid hormone-related protein. Cancer 1991;68:642–7. 29. Allan SG, Lockhart SP, Leonard RCF, Smyth JF. Paraneoplastic hypercalcaemia in ovarian carcinoma. Br Med J 1984;288:1714–5. 30. Lawrence WD, Papas P. Histopathology of ovarian neoplasia. In: Shingleton HM, Fowler WC, Jordan JA, Lawrence WD, editors. Gynaecological oncology: current diagnosis and treatment. London, UK: WB Saunders Company Ltd, 1996:174–202.

The British Journal of Radiology, August 2009