Retrospective study of 19 patients with intramedullary spinal cord metastasis

3.1. Patient characteristics (summarized in Table 1) 

Of the 19 patients, 10 were male and nine were female. Mean age at presentation was 56, with a range of 35–75 years. Male patients were older than female patients (mean ages 59 and 50 years respectively). Lung cancer (68%, 13/19) and small cell carcinoma (6/13 patients with lung cancer in our series) was the most frequently observed primary cancer followed by breast carcinoma (16%, 3/19), digestive carcinoma (colorectal and oesophagus) (11%, 2/19) and thyroid carcinoma (5.5%, 1/19). ISCM was predominantly located in the conus medullaris and cauda equina (58%, 11/19) followed by thoracic and cervical cord. The time interval from the diagnosis of primary cancer and development of ISCM was variable (0–11 years) with a median of 21 months. In one patient (#14), ISCM was the initial presentation of the malignancy.

Table 1. Summary of cases of intramedullary spinal cord metastases

	Patient	Gender, age (yr)	Primary cancer	Time since primary cancer diagnosis	Level	Other Mets	Symptoms	Ambul grade	SX	RT	F-U, Ambul grade	F-U (mo)	Miscellaneous
	1	M, 43	Lung, epidermoid	2 yr	C3–C6	Brain	Paraparesis	III	Y	N	Died (2 mo), III	2	Biopsy, PT
	2	M, 68	Lung, adeno	1 yr	T5	No	Paraparesis, Sphinct dis.	III	N	N	Died (4 days), III	0, 4 days	Abstention
	3	M, 61	Oesophagus	1 mo	C4	Brain	Quadriparesis, Sphinct dis.	III	Y	Y	Slight improvement, Died (3 mo), II	3	Post op RT
	4	M, 50	Lung, small cell	11 mo	Conus	No	Rachialgia, Sciatica dysuria	II	Y	Y	Improvement, Died (6 mo), I	10	Post op RT, ChT
	5	F, 39	Breast	17 mo	T5 and T8	No	Brown-Sequard	II	Y	N	Worsening (paraparesis), Died (2 mo), III	2	Development of multiple brain metastases 1 month after surgery
	6	F, 47	Lung, adeno	18 mo	C2	Brain (SX)	Rachialgia, Paraparesis, Pyramidal sd, Sensibility dis, Sphinct dis.	III	Y	Y	Improvement, Able to walk, Died (8 mo), II	8	(1) Palliative RT then (2) SX indicated for neurological deterioration. Development of new brain mets 6 months after cord surgery
	7	F, 39	Lung, small cell	1 yr	Conus	No	Pyramidal sd, Sensinility dis, Sphinct dis.	I	N	N	Died (4 mo), III	4	Declined any treatment
	8	F, 58	Thyroid	11 yr	T12	Lung (SX), Brain (SX)	Acute paraplegia following fall	III	Y	N	No improvement, Died (4 mo), III	4	Development of multiple brain metastases, 3 months after SX
	9	M, 79	Colon	2 yr	Conus	Liver	Paraparesis, Sphinct dis.	III	Y	Y	Slight improvement, Died (5 mo), II	5	SX requested by patient, Post op RT
	10	M, 66	Lung, adeno	1 yr	Conus	Brain, Liver	Sciatica, Sphinct dis., Cauda equina syndrome	II	N	N	Died (1 mo), III	1	Abstention
	11	M, 57	Lung, adeno	3 mo	Conus	No	Paraparesis	III	Y	Y	Worsening, Died (4 mo), III	4	Post op RT, ChT
	12	F, 56	Lung, small cell	1 yr	T4	No	Paraparesis, Urinary incontinence	III	Y	Y	Worsening, Died (5 mo), III	5	Post op RT
	13	M, 49	Lung	18 mo	(1) Conus exophytic lesion, (2) multiple small mets of cauda equina	Orbit	Severe paraparesis, Radicular pain, Sphinct dis.	III	Y	Y	Improvement, Able to walk then died (2 mo), II	2	SX (excision of conus met), Post op RT
	14	F, 63	Breast	3 mo	Conus	No	Walking disab., Sensory disorders, Urinary incontinence constipation	II	Y	Y	Significant improvement, Able to walk, Alive, I	24, Alive	Post op RT, ChT followed by breast surgery 6 months later
	15	M, 66	Lung, adeno	6 mo	T9	No	Severe paraparesis	III	N	N	Died 2 weeks, III	0.5	Abstention, PT
	16	M, 55	Lung, epidermoid	2 yr	Conus and cauda equina	No	Cauda equina syndrome rapidly completed	III	Y	Y	Improvement, Died (11 mo), III	11	Post op RT
	17	F, 68	Lung, small cell	2 yr	Conus	No	Sensory disorders, Walking disab.	II	Y	Y	Improvement, alive, I	15, Alive	Post op RT, Breast cancer 13 years ago, Melanoma 9 years ago
	18	F, 35	Breast	4 yr	Conus	Bone, multiple	None	0	N	N	Alive, 0	6, Alive	ChT
	19	F, 65	Lung, small cell	20 mo	Conus and cauda equina	No	Sciatica, Paraparesis	III	Y	Y	Improvement, alive, I	4, Alive	Post op RT

Abbreviations: Mets, metastases; SX, surgery; RT, radiotherapy; ChT, chemotherapy; PT, palliative therapy; Ambul grade, ambulatory grade; F-U, follow-up; sphinct dis., sphincter disorders; walking disab., walking disability; sd, syndrome; epidermoid, epidermoid carcinoma; adeno, adenocarcinoma; yr, year(s); mo, month(s); Y, Yes; N, No.

One patient (patient #18) was symptom-free at presentation, the ISCM having been diagnosed via MRI performed to assess the extension of known vertebral metastases. Twelve patients (63%) were non-ambulatory (grade III) and six (31%) had slight functional impairment (5 patients in grade II and 1 patient in grade I). Due to the significant number of ISCM involving conus medullaris and cauda equina, 68% patients manifested urinary and/or anal disorders. Three patients presented with Brown-Sequard syndrome, defined as markedly asymmetric paraparesis with unilateral or more pronounced hypalgesia on the less paretic side. In our series, the mean duration of onset of neurological symptoms was 16 days (1 day to 5 months).

Nine patients (47%) had systemic and/or CNS metastases at the time of presentation. Among them, six had one or more brain metastases. Three patients developed additional brain metastases within a few months following spinal cord surgery. For 11 patients (58%), ISCM was the first metastasis that was diagnosed in the course of the cancer.

3.2. MR imaging findings 

In all patients, MRI with T1 and T2 imaging sequences and gadolinium enhancement was performed. A well-defined nodule with marked enhancement was consistently observed (Fig. 1, Fig. 2). In all cases except 3, the spinal cord metastasis was unique. One patient (#5) had two ISCM located at levels T5 and T8. In two cases (#13 and #19), the ISCM had developed in the conus and was associated with multiple small tumors disseminated along the cauda equina (Fig. 3). MRI clearly showed leptomenigeal involvement in seven cases and the presence of intratumoral cysts in 10 cases.

View Large Image Download to PowerPoint

Fig. 1. Case 6. Intramedullary spinal cord metastasis of lung origin. Sagittal T2-weighted image showing a small intramedullary tumor located at C2–C3 level. Note the presence of hyperintensity above and below the tumor corresponding to the surrounding edema.

View Large Image Download to PowerPoint

Fig. 2. Case 14. Intramedullary spinal cord metastasis of breast origin. Pre- (A) and postoperative (B) MRI. Sagittal T1-weighted images with gadolinium enhancement (fat saturation) showing the complete resection of the ISCM located within the conus medullaris. The postoperative MRI was performed 1 year following surgery. Limits of the laminectomy are indicated (arrows). This patient is still alive 24 months following surgery and is able to walk without aid.

View Large Image Download to PowerPoint

Fig. 3. Case 19. Intramedullary spinal cord metastasis of lung origin. Sagittal T1-weighted image with gadolinium enhancement (fat saturation) showing an intramedullary tumor located within the conus medullaris. Note the leptomeningeal involvement and the presence of two small tumors involving the cauda equina (arrows).

3.3. Patient management 

Thirteen patients (68%) underwent microsurgical excision of the ISCM. Tumor removal was complete in 11 patients and incomplete in two patients (#13 and #19). In one patient (#1), a surgical biopsy of ISCM was performed. Surgical abstention was preferred in five patients. Reasons that motivated surgical abstention were (1) poor general status and diffuse metastatic disease in three patients; (2) refusal of surgery by one patient; (3) asymptomatic ISCM in one patient.

Eleven patients (58%) received external beam radiotherapy with total doses comprised between 20 and 30 grays. Following surgery, 10/13 patients underwent RT focused on the operative bed. Two operated patients (#5 and #8) did not receive RT due to rapid deterioration of their general status. One patient (#6) was initially treated with palliative RT and then underwent surgery when neurological deterioration occurred.

3.4. Follow-up 

The overall median survival was 6.1 months (range: 4 days to 24 months). Median survival was significantly longer in surgically treated patients than in conservatively treated patients (7.4 months vs. 2.6 months, p<0.03). The Kaplan–Meier survival curves are shown in Fig. 4.

View Large Image Download to PowerPoint

Fig. 4. Graph showing Kaplan–Meier estimates of overall survival of operated (black line) and non-operated (dotted line) patients with intramedullary spinal cord metastases in months. Median survival following surgery was 7.4 months in the surgical group and 2.6 months in the non-surgical group (Log-rank test: p<0.03). Note that four patients are still alive (3 in the surgical group, 1 in the non-surgical group).

4.1. Epidemiology 

Mean age at presentation was 55 years and men were slightly more affected than women [29]. In an autopsy series, Connolly reported that ISCM affected the cervical, thoracic and lumbar cord equally [7]. Kalayci et al. [29] and other authors [16], [50], [60] reported the cervical cord as being the most common site (62%). In our series, the most common site was the conus medullaris (58%, 11/19). ISCM were most often single. One-third of patients with ISCM had concomitant brain metastases (35%), and one-quarter had leptomeningeal carcinomatosis [29]. Most ISCM patients had systemic metastases at the time of diagnosis [7], [29], [42]. In our series, 42% of patients had systemic and/or CNS metastases at the time of ISCM diagnosis. The average time between the patients’ initially diagnosed tumor and ISCM occurrence varied from 13 months [45] to 17 months [52]. In our series, the average time was 21 months, reflecting the longer survival currently observed in cancer patients.

Patients with ISCM may exceptionally be asymptomatic (1%) [29]. In our series, only one patient (#18) was asymptomatic. In Kalayci’s review, symptoms were present in 99% of patients including weakness (91%), sensory loss (79%), sphincter dysfunction (60%) and pain [29].

Like SEM, ISCM can produce myelopathy and back pain. Pain tends to be a little less ubiquitous and less severe than with SEM. Another clinical feature that could suggest the presence of ISCM is Brown-Sequard syndrome, or a strong asymmetry within a myelopathy [2], [29]. This has been seen in almost half of ISCM patients and only 3% of patients with SEM [52]. In our series, 15% of patients presented with Brown-Sequard syndrome. Rapid progression of symptoms distinguishes ISCM from primary intramedullary tumors such as glioma, which typically present with a slower progression of symptoms. In three-fourths of previously reported cases, the time from the onset of neurological symptoms to the development of the full neurological deficit varied between 1 [22], [28], [52] and 2 months [29]. In our series, the mean duration of onset was shorter (16 days).

ISCM has been reported as being the initial presentation of malignancy in 25% of cases [13], [29], [52]. In our series, all patients except one had previously detected cancer.

Lung cancer (54%) and especially small cell carcinoma accounted for a majority of cases [2], [6], [25], [29], [36], [41], [43], [45], [50], [52], [63] with breast carcinoma [29], [35], [45] (11%), melanoma (8%), lymphoma [38] (4%), renal cell carcinoma [3], [12], [15], [48], [53] (9%) as other common culprits. In our series, the primary cancer was lung cancer in 68% of cases (13/19) and breast cancer in 16% of cases (3/19).

Intramedullary spinal cord metastasis generally has occurred within the setting of widespread metastasis and most patients have died within 3 months [22], [29], [52].

4.2. Imaging aspects 

When a spinal lesion is suspected in a patient with a history of malignancy, MRI with gadolinium enhancement should be performed to detect ISCM as well as epidural tumor [10], [18], [30], [37], [46], [54], [57].

MR scan readily differentiates ISCM from SEM, revealing parenchymal circumscribed contrast enhancement with a larger surrounding area of T2 signal, indicative of edema. On T1-weighted images, tumors are homogeneous with areas with low signal intensity as well as isointensity. In a few cases, there is a subtle area of increased intensity. Following injection of gadolinium-DTPA, a well-defined nodule becomes visible, whereas the surrounding tissue remains unchanged, consistent with peritumoral edema. In several cases, an intra tumoral cyst with lack of enhancement on T2-weighted images is clearly apparent following injection of gadolinium.

Several authors have demonstrated that positron emission tomography (PET) particularly with fluorine-18-fluorodeoxyglucose enables examination of the entire spinal cord and thus permits the detection of ISCM, especially those of renal or lung origin [27], [34], [48].

4.3. Management and factors influencing survival 

Management (surgery, radiotherapy or abstention) of ISCM is neither simple nor predefined and should be decided case by case. Patient wishes must also be taken into account.

Treatment of ISCM generally consists of fractioned RT, which usually arrests tumor growth and prevents further neurological deficit [51], [52]. It must be noted, however, that only radiosensitive metastases such as small cell lung carcinoma, breast cancer or lymphoma might respond to RT [6], [36], [41], [50], [56], [61]. As mentioned above, ISCM origin is most frequently found within these three malignancies. Previous reports favoring RT are somewhat biased as their clinical elements mainly consist of these radiosensitive tumors [36], [41], [50]. Due to the rarity of ISCM, no controlled studies comparing surgery and RT when administered alone or in combination are available and no recommendation can be strongly made. In our institution, the trend is to manage ISCM like brain metastases and as a result, to propose postoperative RT (20–30 grays) focused on the operative bed. External RT beam treatment is considered as being more effective in cases lacking severe neurological impairment [61]. Once paraplegia sets in, the response to local radiation treatment is poor [29], [36], [52].

Surgery is rarely indicated and most surgeons are reluctant to perform aggressive surgery, given the inherent risks of operating on the spinal cord and the habitual state of advanced cancer elsewhere in these patients. Patients in poor general health and having severe neurological deficits and multiple metastases represent obvious contra-indications to surgery. In our series, three patients (#3, #5, #13) underwent surgery and subsequently survived only a short period of time (<3 months). Three other ones (#8, #11, #12) remained paraplegic following surgery and died 4–5 months later. Retrospectively, it could appear that surgery may not have been a viable option for these patients. Surgery should not be recommended in either cases of complete paraplegia or in asymptomatic patients. No improvement can be expected following surgery when paraplegia is complete. Conversely, removal of ISCM carries the risk of impairing an asymptomatic patient.

However, in selected cases (patients with well-controlled radio-resistant primary tumors and solitary, intramedullary metastases), surgery should receive more consideration and there are some data demonstrating that complete resection when possible improves both quality of life and survival [7], [11], [16], [17], [21], [23], [29], [44], [55], [56]. In 2004, Kalayci reviewed 138 previously published and well-documented cases [29]. Thirty-three patients underwent surgery. Postoperative neurological improvement was observed in 66% of patients. Median survival was 9.4 months when patients underwent surgery and 5 months when conservative treatment was adopted [29]. In 2005, Gasser et al. [21] published a series of 13 surgically treated patients. Median overall survival was 31 weeks (7.2 months) and median progression-free survival was 13 weeks. In our series, survival was significantly longer when surgery had been performed (7.4 months vs. 2.6 months). Most importantly, the neurological status was improved in 7/13 of the operated patients. The longest survivors (patients #11, #14 and #17) all underwent surgery and two patients (#14 and #17) are currently surviving with good general and neurological status respectively 24 and 15 months following the resection of the ISCM. In both of these patients, the postoperative MRI demonstrated a complete excision of the ISCM and the malignancy is presently under control.

Obviously, due to retrospective nature of the study with its related selection bias, one should be cautious in analyzing the results. The better survival rate in surgical groups may be selection biased related. Patients who have wide systemic disease burden are likely to be unable to tolerate surgery. On the other hand, those with good systemic disease control and with good clinical grade are likely to receive surgery and continue to better survival advantage due to the lack of systemic disease.

The purpose of surgery is to allow patients the best autonomy possible by improving their neurological status. Thus, surgery should be proposed as soon as possible to patients having neurological deficits and experiencing rapidly progressive deterioration of neurological function. Because surgical results are better in patients having slight neurological deficits, ICSM should be diagnosed as early as possible by performing MRI of the whole spinal cord with gadolinium enhancement. Results are directly related to preoperative status. Although 75% of patients in grade II improved to grade I, only 50% of patients in grade II improved to grade II and none to grade I.

Advances in the field of spinal cord microsurgery allow the resection of spinal cord tumors with an acceptable morbidity rate. Intraoperative localization of the tumor is facilitated by the use of ultrasound echography thus limiting the extent of myelotomy. Dissection tools (cavitron ultrasonic surgical aspirator, laser) may be helpful in removing spinal cord tumors. Intraoperative monitoring of somato sensory evoked potentials is mandatory to make this surgery safer and is currently available in most neurosurgical units. Thanks to these advances, surgery of spinal cord tumors has considerably evolved in our institution during the last decade. Surgical approach is more precise, more limited, less invasive and allows complete excision of well-delimited cord tumors like ISCM. In our series, the last four patients underwent surgery with the assistance of the most recent tools described above. None experienced postoperative neurological impairment.

Technically, ISCM without leptomeningeal involvement is well suited to surgery. Like intracranial metastases, ISCM are often well-defined tumors and subject to gross total resection. In most cases, a satisfactory cleavage plane exists between the metastastic tumor and the spinal cord parenchyma, thus allowing complete excision of the tumor with minimal parenchymal traction. In case of leptomeningeal involvement, radical excision is impossible to obtain and it is preferable to limit removal to the well-defined portion of the tumor.

Gasser et al. [21] found that histology was the major determinant of patient survival. Survival was longer in patients with adenocarcinomas than in those with poorly differentiated carcinomas. The possibility of performing radical excision of the ISCM is directly influenced by its histological nature. Poorly differentiated carcinomas and sarcomas proved difficult to dissect due to the lack of a clear cleavage plane. In these cases, these authors recommend biopsy, decompressive laminectomy and adjuvant therapy, representing a valuable alternative to radical resection which can produce devastating results when considering the comparatively short span of survival in this subgroup of patients [21].

Due to the small size of the previously published series, there are no available data regarding other clinical prognostic factors. In our series, we found that neurological status and nature of primary cancer, as well as the presence of systemic and/or CNS metastases could influence median survival but differences were not statistically significant.