It may be true that a more metabolically active tumour may have a higher proliferative capacity, have the potential of acquiring tumour volume at a higher rate and possibly be more locally aggressive compared to a less metabolically active tumour. However, there are host-related mechanisms, which are independent of tumour metabolism and not fully understood, that influence tumour spread, metastasis and survival.
In order to survive and metastasize, it is hypothesized that tumour cells from the primary site would be required to come in contact with the terminal lymphatics, survive in the lymphatic space, invade lymph nodes where they would need to evade the host immune response, continue to proliferate and be selected for clonal expansion to replace lymph nodes with metastases. These cells also need to come in contact with intertissular lymphatic-venous anastomoses [28], pass in to circulation, survive and find a suitable tissue to colonize and form distant metastases.
When a matrix of invasiveness and local growth is observed by MRI and the distribution of viable tumour colonies is detected by PET imaging, a complete snap shot of tumour and host interaction is assessable and accurate prediction of prognosis becomes possible. A single value, (or a range of values simply based on metabolic capacity of the primary tumour devoid of any host interaction parameters), such as SUVmax alone, would only enable one to know the potential of the tumour to exhibit a more or less aggressive behavior, rather than actually predicting the likelihood of recurrence. For example, in uterine cancers, it is known that serous and clear cell histologies have a poorer prognosis compared to endometerioid histologies. However, the overall prognosis is further modified by other prognostic factors, such as the presence or absence of Lymphovascular Space Invasion (LVSI) or lymph node metastasis. In fact, patients with clear cell and endometrioid histologies that are lymph node negative and do not have LVSI, were both found to have <8% recurrence rates [29]. However LVSI positive endometrioid histology will still have worse prognosis than LVSI and node negative clear cell or serous cancers of uterus.
Therefore, just as “bad” histology on its own is not sufficient to predict prognosis, we believe that an increased SUVmax (a sign of increased metabolic activity) in isolation is likewise insufficient as a reliable predictor of treatment outcome.
In a study by Kidd et al., [30] a PET-based nomogram was used which included tumour volume, SUVmax of primary tumour and presence of and level of nodal involvement. The authors found that the most significant poor prognostic factor was the detection of lymph node metastasis by PET imaging and the level of nodes, rather than their SUVmax. SUVmax value was only marginally better than tumour volume for predicting Relapse Free Survival (RFS). The HR for OS in locally advance cervix cancer was 0.6, 1.12, 1.40 and 3.24 in node negative, pelvic, common iliac and para-aortic node positive patients respectively [25], in comparison, HR for OS was of 1.009 for SUVmax of >14 for primary tumour [30]. Perhaps the prognostic significance of SUVmax of the primary tumour could be better explored in PET node negative patients for it to have a practical application.
We believe several aspects related to SUVmax need to be explored before it could become a practical and reproducible prognostic parameter useful in the management of cervix cancer patients.
- Is SUVmax a continuous variable or it can be subdivided in quartiles or even dichotomized?
- What is the best way of integrating the SUVmax of Lymph Nodes (SUVLN) in the presence of nodal metastases in the overall matrix of a SUVmax based prognostic tool?
- Are there different implications on the interpretation of the prognostic potential of SUVmax (primary tumour) in node negative and node positive patients? Should there be separate SUVmax based prognostic tools?
- Since SUVmax is measured in relation to background uptake, for example, in liver, where this background uptake could vary from patient to patient, is there a need for a universal correction factor?
We have more than 300 patients of cervix cancer treated with curative intent between 2002 and 2010 who underwent pretreatment PET and MRI studies. The details of the pretreatment assessment, histopathological data, treatment and follow-up have been prospectively collected and recorded in our institutional database. We are currently testing the hypothesis that there would be a trend of higher SUVmax with the lymph node metastases but survival will relate the level of nodal metastases and only in node negative patients SUVmax may have a significant prognostic role in stratifying patients over tumour volume and the FIGO stage of the disease.
As a pilot study to test this hypothesis, we randomly selected cervix cancer patients from 4 different categories thought to increasingly have poor prognosis. (n=16)
- Small primary and negative nodes
- Small primary with positive nodes
- Large primary with negative nodes
- Large primary with positive nodes
The initial disease parameters upon diagnosis, related known prognostic factors, as well as subsequent relapse time line, the site of relapses, disease status and at last follow-up is shown in table 1. The pre-treatment SUVmax of the primary tumour in relation to their other parameters and subsequently outcome is shown. In this cohort of patients, there were no primary site or pelvic nodal failures. Only 1 out of 7 node negative patients have relapsed, of whom the pre-treatment SUVmax of primary tumour ranged from 4.2 - 17.5. The one who failed had SUVmax 14.4. Out of the 9 node positive patients relapse (SUVmax ranged from 6.7 - 46), 4 relapsed (SUVmax were 9.3, 21.9, 26.7 and 46 respectively). It is interesting to note that patient No 8 who had a small tumour with nodal involvement involving upper PA node, had SUVmax of 9.3, relapsed in supraclavicular nodes and lung and mediastinum and received high dose palliative radiotherapy to these sites is alive with disease present only in lung at >6 years since her initial diagnosis.
Patient |
FIGO |
SUV |
MRIVol |
No + nodes |
Highest Involved node |
SUV Max |
Relapse & Follow up |
Stage |
max |
(LN) |
|
(Primary) |
|
Site |
Status |
Days Survived |
1* |
2A |
4.2 |
1 |
0 |
|
|
|
Dead (other) |
393 |
2* |
1B |
2.5 |
4 |
0 |
|
|
|
Alive |
2426 |
3* |
2B |
6.7 |
15 |
0 |
|
|
|
Alive |
2325 |
4* |
3A |
9.1 |
|
0 |
|
|
|
Alive |
2244 |
5# |
1B |
6.7 |
1 |
2 |
Inf PA |
3.1 |
|
Alive |
1384 |
6# |
1B |
10.6 |
6 |
4 |
Common Iliac |
13 |
|
Alive |
2014 |
7# |
2A |
24.4 |
6 |
4 |
Common Iliac |
3.8 |
|
Alive |
1568 |
8# |
2A |
9.3 |
27 |
4 |
upper PA |
4.9 |
|
Alive |
1842 |
9$ |
1B |
15.6 |
52 |
0 |
|
|
|
Dead (other) |
805 |
10$ |
2B |
17.5 |
69 |
0 |
|
|
|
Alive |
2041 |
11$ |
1B |
14.5 |
100 |
0 |
|
|
PA, D |
Dead (tumour) |
513 |
12@ |
3B |
26.7 |
84 |
2 |
Pelvic |
15.5 |
PA, SC, D |
Dead (Tumour) |
1079 |
13@ |
3B |
21.9 |
201 |
2 |
Pelvic |
10.7 |
D |
Dead |
1039 |
14@ |
2B |
46 |
238 |
4 |
Common Iliac |
33.5 |
D |
Dead |
1008 |
15@ |
2B |
15.2 |
291 |
3 |
Inf PA |
6.9 |
|
Alive |
1658 |
16@ |
4A |
19.2 |
293 |
4 |
Common Iliac |
12.5 |
|
Alive |
1464 |
Table 1: Patient and disease characteristics, SUVmax and subsequent follow up following definitive chemo-irradiation.
PA= Para Aortic; Inf = Inferior; SC = Supraclavicular; D = Distant; P = Primary; N = Node
* = small primary, node negative; #= small primary, node positive; $ = large primary, node negative; @ = large primary, node positive
To our knowledge, most of previously published studies have attempted to relate the SUVmax of primary tumour with survival by utilizing various methods of obtaining a cutoff value that may fit their data. However, in addition to the limitation of small sample sizes and variable tumour types, no uniform value has emerged. The independent and clinically useful prognostic potential of SUVmax still remains undecided.