Therapy monitoring of small intestinal neuroendocrine tumors with CT; quantification of tumor attenuation and contrast-enhancement

(1)

Therapy monitoring of small intestinal neuroendocrine tumors with CT; quantification of tumor attenuation and

contrast-enhancement

By: Bazyli Grosz

Supervisor: Anders Sundin

14/1-2017

(2)

1 Abbreviations and definitions

NET: Neuroendocrine tumor

SI-NET: Small intestinal NET

Ki-67 index: Immunohistochemical marker of cell proliferation. Higher Ki-67 index means higher proliferation in the tumor and is often associated negatively with the clinical course of a malignancy.

CT: Computed tomography

HU: Hounsfield Units, a measure of X-ray attenuation or “tissue density”.

Native phase: CT without contrast enhancement.

Late arterial phase: Approximately 35-40 seconds after contrast injection start.

Structures supplied by arteries will show optimal contrast- enhancement and be optimally visualized.

Venous phase: About 60-90 seconds after contrast injection start. The parenchymal phase in which the liver parenchyma is best contrast-enhanced when the contrast medium has been delivered to the liver through the portal vein.

Hypervascular metastasis: Well vascularized metastases, that are best depicted on CT in the late-arterial phase.

Hypovascular metastasis: Poorly vascularized metastases, often with extensive necrotic components, that are best depicted on CT in the venous phase.

PRRT: Peptide Receptor Radionuclide Therapy, a molecular targeted therapy performed by i.v. infusions of a radiopeptide

consisting of a somatostatin analog coupled to a beta-emitting radionuclide such as Lutetium-177 (¹⁷⁷Lu).

PFS: Progression free survival

(4)

2 Abstract

Neuroendocrine tumors (NET) are a heterogeneous group of tumors that differ in many aspects from the more common malignancies not least when it comes to imaging. CT is the standard radiological modality for clinical work-up and surveillance of NETs. Yet there is no optimal way for monitoring of therapy response in NETs, neither with CT nor any other modality. The objective of this research project was to examine whether attenuation measurements of liver metastases before and after peptide receptor radionuclide therapy (PRRT) with ¹⁷⁷Lu-DOTATATE can be useful for monitoring of therapy response in patients with small-intestinal neuroendocrine tumors (SI-NET). In this retrospective observational study, measurements of late arterial phase attenuation and contrast-enhancement in liver metastases before and after PRRT were performed in 42 patients out of the original cohort of 233 SI-NET patients undergoing PRRT. Student’s T-test was used to examine whether there was a mean difference between measurements after the treatment compared to those at baseline. Kaplan-Meier analyses were performed to correlate these results to the clinical outcome (Progression free survival). There was no clinically significant change in late arterial phase attenuation or contrast-enhancement in the liver metastases after treatment and the results did not correlate to clinical outcome. More studies are needed in order to find better ways for monitoring therapy in NETs.

(5)

3 Populärvetenskaplig sammanfattning på svenska

Detta projektarbete har som syfte att testa en alternativ metod för radiologisk utvärdering av behandling för neuroendokrina tumörer (NET) som utgår från tunntarmen. Vid spridd sjukdom brukar man i första hand behandla tunntarms-NETs med somatostatinanaloger men när denna behandling börjar svikta går man över till så kallad ”peptid receptor radionuklid terapi” (PRRT) där man kopplar på en beta-strålande radionuklid till somatostatinanalogen, vanligen Lutetium-177 kopplad till octreotat (¹⁷⁷Lu-DOTATATE), och på så sätt åstadkoms en effektiv målsökande intern strålbehandling av tumören. Behandlingens effektivitet åtföljs dock av en hög kostnad och den kan ge allvarliga strålningsrelaterade biverkningar. PRRT har visat sig ge bättre överlevnad hos hela gruppen patienter med tunntarms-NET men är svår att utvärdera för den enskilda patienten i ett tidigt skede. Gängse radiologiska

utvärderingsmetoder, där man mäter tumörstorlek, har visat sig vara bristfälliga för NETs. För andra tumörformer, som har vissa likheter med NETs, har man på senare år framgångsrikt använt sig av tumörattenuering (mått på tumörens röntgentäthet) vid datortomografi (DT, Eng. computed tomography, CT) som mått för att utvärdera behandling men man har ännu inte testat det på tunntarms-NETs.

Vi har därför i detta arbete undersökt om mätning av attenuering och kontrastuppladdning före och efter behandling i levermetastaser från patienter med tunntarms-NET kan vara av värde för utvärdering av PRRT. På grund av strikta inklusionskriterier har vi endast kunnat inkludera 42 patienter i studien. Hos dessa har vi påvisat en minskning i metastasernas attenuering och kontrastuppladdning efter behandling men skillnaderna var för små för att vara signifikanta. Vi har även undersökt om skillnaden i attenuering/kontrastuppladdning korrelerar till hur det går för patienterna (tid till radiologisk progress) men har inte funnit något statistiskt samband därmed. Därför måste vi dra slutsatsen att mätning av

tumörattenuering och kontrastuppladdning inte är av värde för utvärdering av PRRT för tunntarms-NET, baserat på detta urval av patienter. Fler studier behövs inom området.

(6)

4 Introduction

4.1 Neuroendocrine Tumors

Neuroendocrine tumors (NET) are a heterogeneous group of tumors that arise from

neuroendocrine cells of the diffuse endocrine system, which is dispersed throughout the body.

They constitute a relatively rare group of malignancies but their reported incidence has increased over the last years (1,2). As a result of improvements in diagnostics and treatment the NET-patients today live considerably longer, particularly when treated at specialized centers, which means that the prevalence is considerably higher and corresponds to that for esophageal-gastric cancer and pancreatic cancer (3). NETs can be classified in various ways.

One way of classifying NETs relates to its site of origin. Small-intestine, lungs-bronchi and colon-rectum are the most common locations. This study will focus on NETs originating in the small intestine (SI-NETs), previously known as midgut-carcinoids, which constitute about 25% of all NETs. (4) NETs may produce peptides or hormones, which can cause hormonal symptoms, and are then called ”functioning” NETs, as opposed to ”non-functioning” NETs that do not give rise to such symptoms, even though hormonal production can be detected biochemically (5). The tumor biology of NETs differs from that of the malignancies in general oncology in the sense that they rarely present as clinically aggressive tumors.

Amongst the NETs that are regarded as malignant, the majority still consists of tumors with well-differentiated cells and with a low proportion of cells in the proliferative phase. The third way of classifying NETs, the tumor grade (G), is defined through immunohistochemical staining of tumor cells in order to quantify the proportion (%) of cells in mitosis (Ki-67 index), G1 >2%, G2 2-20% and G3 >20% (6). S-I NETs generally have low proliferative activity often with ki-67 from less than one to a few percent.

NET Symptoms are usually vague, such as abdominal discomfort and pain. Consequently, in many cases it takes years to establish the diagnosis because of both “patient’s” and “doctor’s delay” and approximately 50% of the small-intestine NET-patients present with disseminated disease at diagnosis with equal parts regional spread (25%) and distant metastases (25%) (5).

Also, many small-intestinal NETs are asymptomatic at presentation, due to their slow growth, and are only discovered accidentally (7). SI-NETs usually present a characteristic metastatic progression pattern. The primary tumor is usually small, often less than 1 cm, and it is not uncommon with multiple primaries. Lymphatic spread is frequent and usually leads to the development of a mesenteric metastasis that on radiological imaging often has a characteristic

(7)

appearance with calcifications and surrounding fibrosis caused by the release of serotonin from the metastasis. The mesenteric metastasis can lead to intestinal obstruction by adhesion of intestinal loops or luminal stricture as well as venous ischemia due to kinking of the mesenteric veins (3,8,9). Furthermore, lymphatic spread can give rise to lymph node metastases in the retroperitoneal space, the mediastinum and sometimes in the neck (10).

Hematogenous spread of SI-NETs occurs mainly to the liver and in cases with high tumor load of functioning NETs this may cause hormonal symptoms with flushing and diarrhea, constituting the so-called carcinoid syndrome. Moreover, serotonin produced by the liver metastases is transported through the hepatic veins and the inferior vena cava to the right heart, which induces fibrosis on the tricuspid and pulmonary valves leading to

stenosis/regurgitation, the so-called carcinoid heart disease (11,12). Hematogenous spread to the bone is common, especially, during the latter stages of the disease (13), while brain and lung metastases are less common (14).

4.2 Computed tomography (CT)

The CT-scanner consists of a tunnel or gantry with a rotating X-ray tube and X-ray detectors positioned on the opposite side of the gantry. With the help of CT-software, the radiation registered at each detector position in the spiral is subtracted from the radiation coming from the X-ray tube rendering a value of the tissues X-ray absorption, the so-called attenuation (Hounsfield Units, HU). The software then reconstructs the data to produce transverse sections of the body’s organs and tissues. Today, by using numerous parallel detectors, very thin (<1mm) transverse images are produced that can be further reformatted into high- resolution sagittal, coronal and 3D-images facilitating image presentation and interpretation.

CT is the standard radiological modality for clinical work-up and surveillance of NETs (3).

4.3 Liver metastases in NET patients

The blood supply of the normal liver parenchyma is mainly by the portal vein (approximately 75-80%) and only a minor fraction of the blood is supplied by the hepatic artery

(approximately 20-25%). Malignant tumors in the liver recruit their blood vessels from the branches of the hepatic artery and never from the portal-venous vessels. (5)

As a rule, malignant tumors grow rapidly and are poorly vascularized. Poorly-vascularized metastases are referred to as “hypovascular”. In general oncology, liver metastases are therefore best depicted on CT during intravenous contrast-enhancement in the venous phase, i.e. when the normal liver parenchyma is fully contrast-enhanced by the contrast medium delivered through the portal vein whereas only a minor fraction of the contrast medium is

(8)

supplied to the metastases via the hepatic artery. This is further accentuated by the fact that the metastases are generally poorly vascularized. Consequently, the hypovascular metastases will, due to their low contrast-enhancement, appear darker (low attenuating) than the

surrounding well perfused and contrast-enhanced bright (high attenuating) normal liver (16).

By contrast, most NETs, unlike the majority of malignancies, have low proliferation and grow slowly. NET metastases, therefore, usually become better vascularized than their counterparts from the more common cancer types. Hypervascular NET liver metastases are therefore frequent especially from pancreatic NETs. These well-vascularized liver metastases will be best depicted on CT in the so-called late arterial phase, when the contrast medium has reached the main arteries and further into the small hepatic arteries and tumor vessels. The tumors are at this time point contrast-enhanced, high attenuating (bright) while the surrounding normal liver is virtually unenhanced, low attenuating (dark), since the contrast medium has not yet reached the normal liver parenchyma through the portal vein. Although the presence of hypervascular liver metastases in NETs is frequent as compared to the more common cancers this is, however, not a rule and poorly vascularized NET metastases are not infrequent (5).

Thus, NETs present a diverse pattern of metastatic spread with well-vascularized lesions or poorly vascularized lesions and often a combination thereof. Therefore, when imaging NETs, the liver and pancreas should always be examined before and after intravenous contrast infusion in the late arterial phase as well as in the venous phase in order to be able to depict well-vascularized lesions as well as poorly vascularized lesions (17,18).

4.4 Therapy

Surgical resection is the only curative treatment for NETs. In metastatic disease, the choice of systemic therapy depends on the site of origin and tumor grade (G1-G3). SI-NETs, almost always constitute G1 and low G2-tumors and are generally treated with monthly injections of long acting somatostatin analogs. When this fails, alpha-interferon can be added.

Locoregional therapies that are used in progressive disease include radiofrequency ablation, microwave ablation and intra-arterial embolization or chemoembolization of liver metastases (5). In NETs other than SI-NETs, molecular targeted therapies are available when first line therapy fails such as mTOR and tyrosine kinase inhibitors (19,20). Further, targeted

radionuclide therapy is increasingly being used. It is performed by conjugating a somatostatin analog to a beta emitting radionuclide. In Sweden, Lutetium-177 in the form of 177Lu-

(9)

DOTA-octreotat (177Lu-DOTATATE) is used. This so called peptide receptor radionuclide therapy (PRRT) is administered as an i.v. infusion every 6-8 weeks (21,22). The length of the treatment is determined by the number of treatment cycles it takes to reach the maximal radiation dose to the bone marrow (2Gy) and the kidneys (23Gy), which varies between patients. Generally, 3-6 cycles are administered (23). In some cases, the treatment has to be paused and the interval between cycles prolonged because of bone marrow toxicity, or even terminated, due to serious cytopenia or deteriorating clinical status of the patient. PRRT can sometimes be repeated if the disease progresses, usually after 1-2 years. PRRT has shown most effective for treatment of metastatic NET disease (21).

4.5 Evaluating therapy response

In general oncology, RECIST 1.1 (response evaluation criteria in solid tumors) are applied to assess tumor response on morphologic imaging, usually CT/MRI (24). RECIST 1.1 relies on measuring the longest diameters of 2 target lesions per organ, maximum 5 lesions in total.

The sum of the diameters of these target lesions is then used as the comparator between subsequent CT/MRI-examinations to determine the therapy response, which is graded into four categories: progressive disease (PD), stable disease (SD), partial remission (PR) and complete remission (CR). Despite its simplicity and suitability in general oncology, there are some tumor types for which lack of tumor shrinkage, as assessed by RECIST 1.1, does not predict poor outcome. This seems to apply especially in NETs. Even though various NET treatments have shown to increase overall survival, they rarely lead to significant tumor shrinkage. Instead, the treatments usually result in stabilization of the tumor size and in progressing NETs, stabilization is indeed considered as response to the treatment (25).

In general oncology, solid tumors are often highly proliferative and therapy is expected to result in reduction in tumor size. By contrast, the low proliferative nature of NETs may offer an explanation for the lack of reduction in tumor size following treatment. Response to therapy may be associated with changes in tumor physiology (receptor status etc.) or tumor metabolism even in the absence of reduction in tumor size. Many NET treatments, including PRRT, are costly and may be associated with serious side effects. Therefore, it would be desirable to develop a method for early evaluation of therapy response in order to change the treatment strategy in cases when this is ineffective (25). In general oncology, FDG-PET is used for this purpose, particularly for treatments with problematic side effects. Except for high grade (G3) tumors (neuroendocrine cancer, NEC), FDG-PET has not proven useful

(10)

either for lesion detection or therapy monitoring in NETs due to their generally low proliferative nature. In the majority of NETs there is little or no FDG uptake (26). Other functional imaging methods such as 68Ga-DOTATOC PET/CT or F-DOTA-PET/CT have been shown to be superior to conventional radiological methods in visualization of NETs (27,28). Naturally, they also provide useful additional information about tumor functional status. However, there are no conclusive findings regarding their role in monitoring response to therapy and previous reports suggest that they are not suitable for therapy monitoring in SI- NETs, other than early detection of new lesions (25). In summary, there are no robust

methods for early evaluation of NET treatment and consequently there is a need for development of new methods.

4.6 Tumor attenuation as a predictor of therapy effect

The role of tumor attenuation as a predictor of therapy effect have been investigated for other types of tumors. Tumor contrast-enhancement in the venous phase is for example used for monitoring therapy and predicting outcome in gastrointestinal stromal cell tumors (GIST) (29,30), hepatocellular carcinomas (31), adenocarcinomas in the pancreas (32) and renal cell carcinoma (33). It has not yet been studied for SI-NETs.

4.7 Objective

The objective of this study was to examine whether attenuation measurements of liver

metastases before and after PRRT can be useful for monitoring of therapy response in patients with SI-NETs.

4.8 Research questions and hypotheses

Is there a difference in tumor attenuation before and after PRRT in the native phase, in the late arterial phase and in the degree of contrast-enhancement? Can these results be related to the clinical outcome? Are the resulting changes in these respects true treatment effects or the reflection of differences in aortal contrast-enhancement between CT examinations before and after therapy?

(11)

5 Materials and Methods:

5.1 Study design

This is a retrospective observational study in a cohort of patients undergoing CT before, during and after PRRT. No double blinding or randomization was performed. There are no previously published data in this area of research and therefore it is not possible to perform a customary calculation of statistical “power”. The present study must therefore be regarded as exploratory in its nature. Approval was granted from the ethics committee in Uppsala, Sweden (DNR 2011/248).

5.2 Cohort features and selection criteria

Patients were included from the entire cohort of patients with different types of NETs (n=408) that were treated with PRRT at the Nuclear Medicine Department, Akademiska Sjukhuset, Uppsala during the years 2005-2015. In total, there were 233 SI-NET patients of whom 42 met the inclusion criteria (please see Figure 1) and could be included into this study. There were 26 women and 16 men, mean age 64 (range 47-82) years.

5.2.1 Inclusion criteria

Patients undergoing CT of the liver in the native phase (without contrast enhancement) and in the late arterial phase, both before and after PRRT, and who presented with at least one hypervascular liver metastasis at the baseline CT were included.

5.2.2 Exclusion criteria

The selection of patients from the original patient cohort is shown in Figure 1.

The exclusion criteria were:

• No hypervascular liver metastasis at baseline CT.

• One of the following scans missing: Native/late arterial phase before or after treatment.

• Metastases containing extensive calcifications that prevent reliable attenuation measurements.

• Baseline CT performed more than 3 months before the start of PRRT or post-treatment CT performed more than 6 months after the last PRRT-cycle.

(12)

• * This patient received more than twice the regular amount of contrast medium on the baseline CT. Also, the timing of the phase was not optimal.

Figure 1. Flow chart of the patient selection process

5.3 Data collection procedure 5.3.1 Selection of metastases

In the CT examinations before start of therapy (baseline), performed during contrast-

enhancement in the late arterial phase, one liver metastasis per patient was selected according to the following criteria: 1. Level of contrast enhancement: Since the aim of the study is to evaluate changes in contrast enhancement in hypervascular lesions, the lesions with the highest contrast-enhancement were preferably chosen. 2. Tumor size: The largest metastases were favored for better precision in the measurements. 3. Reproducibility: Clearly defined lesions that also were identifiable on follow-up CT were favored.

5.3.2 Outlining regions of interest (ROIs)

The outer contour of each metastasis was defined in the late arterial phase and a

corresponding irregular region of interest (ROI) was defined by using a software based measuring tool and the mean and maximum attenuation (“Hounsfield Units”, HU) of the ROI was registered. A corresponding ROI was positioned in native (pre- contrast) scans. A ROI was also placed in the Aorta, at the level of the celiac trunk, in order to reflect the contrast-

(13)

enhancement of the hepatic artery and the tumor vessels in the metastases. These measurements were repeated in the post treatment CT.

5.4 Calculating attenuation and contrast-enhancement

CT attenuation, measured in Hounsfield units (HU), is a measure of the degree of X-ray attenuation in a given tissue. The CT scanner is calibrated to show 0 HU for water and -1000 HU for air and this interval is divided into 1000 units (-1000 to 0) that is mirrored around the 0-point to show positive attenuation (0-1000). Because the attenuation of cortical bone, and especially of metal implants may exceed 1000 HU, the attenuation scale is extended to about 3600 HU. The native attenuation of the normal liver is approximately 60-70 HU (34). The attenuation of liver metastases varies depending on the composition of tumor components, especially the degree of tumor necrosis. The contrast-enhancement of the metastases was defined as the attenuation in the late-arterial phase minus the attenuation in the native phase.

The corresponding calculation was performed for the contrast-enhancement in the aorta by subtracting the native attenuation from that in the late arterial phase.

5.5 Clinical outcome

Data on clinical outcome was obtained from the digital medical records at Akademiska sjukhuset (Cosmic) and from the radiological reports in the Radiological Information System (RIS) and Picture Archive and Communication System (PACS) (Carestream). The date of tumor progression was determined based on the radiological reports (n=28). In cases where date of progression was unavailable in the radiological reports, all CT examinations from baseline and during the prolonged follow-up were reviewed and assessed according to RECIST 1.1 to determine the date of progression (n=14).

Progression free survival (PFS) is widely accepted as an endpoint for the outcome of therapy in clinical trials and is especially useful when the natural history of the disease being studied is prolonged, as in the case of SI-NETs. PFS is defined as the time period between the start of the treatment and disease progression, or death from any cause. (35)

5.6 Statistical analysis

All statistical analyses were conducted in SPSS V.25.0 software (SPSS Inc, Chicago, Illinois, USA). A Shapiro-Wilk’s test was applied to assess the data for normal distribution. A paired- samples t-test was used to test the first research question: “Is there a difference in tumor attenuation before and after PRRT in the native phase, in the late arterial phase and in the degree of contrast-enhancement?”. Four different Students T-tests were performed to assess

(14)

the attenuation measurements in the metastases. The first tested the mean difference for the contrast-enhancement post treatment and at baseline. The second tested the mean difference between the attenuation in the late arterial phase post treatment and at baseline. The third tested the mean difference for the native attenuation post treatment and at baseline. The fourth tested the mean difference for the maximum attenuation in the late arterial phase post

treatment and at baseline.

The Kaplan-Meier method was conducted to test the second research question: “Can these results be related to the clinical outcome?”. The test explored the relationship between PFS and the change between CT at baseline and after PRRT regarding contrast-enhancement and maximum attenuation in the late arterial phase. Patients were divided into three groups based on their pattern of change and a log rank test was run to determine if there were differences in the survival distributions for the different groups.

To test the third research question “Is the eventual difference in metastatic attenuation attributable to variations in the aortic contrast-enhancement?”, two Pearson’s product- moment correlation tests were applied.

(15)

6 Results:

6.1 Change in tumor attenuation and contrast-enhancement after PRRT (t-test) T-test 1: There were no outliers in the data, as assessed by inspection of a boxplot (Figure 2a) for values greater than 1.5 box-lengths from the edge of the box. The assumption of normality was not violated, as assessed by Shapiro-Wilk’s test (p = .927). The metastatic contrast- enhancement was lower post treatment (Mean = 38 HU, SD = 24) as compared to baseline (Mean= 48 HU, SD = 24), with a significant mean decrease of 9 HU (SD = 25, 95% CI, 1.6 to 17.0, p = 0.019). Figure 3a illustrates the range (-61 to 50 HU) and the distribution of the results. T-test 2: One outlier was detected that was more than 1.5 box-lengths from the edge of the box in a boxplot (Figure 2b). Inspection of the value did not reveal it to be extreme and it was included in the analysis. The assumption of normality was not violated, as assessed by Shapiro-Wilk’s test (p = .231). The metastatic attenuation in the late arterial phase was lower post treatment (Mean = 84 HU, SD = 23.4) than at baseline (Mean = 93 HU, SD = 21), with a significant mean decrease of 10 HU (SD = 23, 95% CI, 2.3 to 16.8, p = 0.011). T-test 3: Five outliers were detected. Three were extreme outliers, more than 3 box-lenghts from the edge of the box (Figure 2c). The assumption of normality was violated as assessed by Shapiro-Wilk’s test (p = .000). The Wilcoxon signed-rank test could have been used in this case but we nonetheless decided to apply the T-test in order to be consistent in the choice of method, and also not expecting significant differences as based on inspection of the data. The native metastatic attenuation post treatment (Mean =45 HU, SD=7) was similar to that at to baseline (Mean =46 HU, SD=9) with a mean 0,3 HU increase (Mean=0,3 HU, SD=11, 95% CI, -3.6 to 3.0, p=0.861). T-test 4: Three outliers were detected that were more than 1.5 box-lengths from the edge of the box in a boxplot (Figure 2d). Inspection of the values did not reveal them to be extreme and they were kept in the analysis. The assumption of normality was violated, as assessed by Shapiro-Wilk’s test (p = .033). The test was nonetheless performed because it is robust regarding data deviations from the normal distribution. Visual inspection of a normal Q-Q plot renders a better impression of the normality of this distribution than the Shapiro- Wilk’s test (Figure 3b). The maximum metastatic attenuation in the late arterial phase was lower post treatment (M = 147 HU, SD = 34) than at baseline (Mean = 164 HU, SD = 47), with a significant decrease of 16 HU (SD=44, 95% CI, 2.6 to 30.1, p=0.021). The results from all four t-tests can be seen in table 1 where negative values of the mean represent a decrease in attenuation and contrast-enhancement after treatment.

(16)

Table 1 Paired samples T-test results.

Figure 2 a) Boxplot from T-test 1. b) Boxplot from T-test 2. c) Boxplot from T-test 3. d) Boxplot from T-test 4.

Figure 3 a) Histogram from T-test 1. b) Q-Q plot of ∆ max attenuation.

Change after treatment in: Mean (HU)

Standard Deviation

95% CI

P-value Lower Upper

1. Contrast enhancement -9 25 -17,0 -1,6 ,019

2. Attenuation late arterial phase -10 23 -16,8 -2,3 ,011

3. Attenuation native phase +0,3 11 -3,6 3,0 ,861

4. Maximum attenuation late

arterial phase -16 44 -30,1 -2,6 ,021

(17)

6.2 Testing the relationship between clinical outcome and changes in metastatic contrast-enhancement and maximum attenuation (Kaplan-Meier Method) Patients that were lost to follow up (n=5) and those who had not progressed as of November 2017 (n=6), corresponding to the end of the study period, were censored. Patients were divided into three groups based on their pattern of change in metastatic contrast-enhancement and that of the maximum attenuation after PRRT: a decrease of more than 10 HU (n=23 for both), an increase of more than 10 HU (n=8 respectively n=11) and a change of 0  10 HU (n=11 respectively n=8), henceforth referred to as “no change”. The Kaplan-Meier analysis compared the impact on the PFS for patients with these three different patterns of change in metastatic contrast-enhancement and maximum attenuation. The Kaplan-Meier analysis was not applied on the results from T-test 2 and 3 because they were inconclusive.

The first analysis tested the relationship between change in contrast-enhancement and clinical outcome (PFS). The number (percentage) of censored cases in the decrease group was 8 (34,8%), in the increase group 2 (25,0%) and in the no-change group 1 (9,1%). Thus, the proportions of censored cases in these three groups were different as was the pattern of censoring (Figure 4a). Median time to progression was 33 months (SE^= 9) in the decrease group, 38 months (SE = 3) in the increase group and 27 months (SE = 3) in the no-change group. A log rank test that was applied and showed that the survival distribution (Figure 5) for the three groups were similar (p = .291).

The second analysis tested the relationship between change in maximum metastatic

attenuation and outcome (PFS). The number (percentage) of censored cases in the decrease group was 6 (26,1%), increase group 2 (18,2%) and in no-change group 3 (37,5%) and was different between groups although the pattern of censoring was fairly similar (Figure 4b).

Median time to progression was 33 months (SE = 9) in the decrease group, 38 months (SE = 8) in the increase group and 23 months (SE = 4) in the no-change group. The survival distributions (Fig. 6) for the three groups were similar (p = .537).

 SE = Standard Error

(18)

Figure 4 a) Censored cases, contrast enhancement b) Censored cases, max attenuation

Figure 5 Kaplan-Meier survival functions of the different groups of change in contrast- enhancement, p = .291.

(19)

Figure 6 Kaplan-Meier survival functions of the different groups of change in maximum attenuation, p = .537.

6.3 Correlation between changes in contrast-enhancement in the metastases and the aorta (Pearson’s correlation coefficient, r)

The first correlation test assessed the relationship between the change in metastatic contrast- enhancement and that of the aorta. Preliminary analysis showed the relationship to be linear with both variables normally distributed, as assessed by Shapiro-Wilk’s test (p > .05). There were six outliers in the scatter plot distributed randomly around the bulk of data points

(Figure 7a). No correlation was found between the differences in contrast-enhancement in the metastases and that of the aorta r = 0.220, between CT at baseline and post PRRT (p = 0.162).

The second correlation test assessed the relationship between the changes in maximum metastatic attenuation in the late arterial phase and that of the aorta. Preliminary analysis showed the relationship to be linear (Figure 7b) while one of the variables (change in

maximum metastatic attenuation) was not normally distributed, as assessed by Shapiro-Wilk’s test (p = .033). The test was nevertheless applied because it is robust regarding deviations from normality. Again, the visual inspection of a normal Q-Q plot of this variable (Figure 3b) provided a much better impression of the normality of this distribution. There were three outliers in the scatter plot. No correlation was found between the difference in maximum

(20)

attenuation in the metastases and that of the aorta (r = 0,101, p = 0,528). The correlation of the results of T-test 2 and T-test 3 to the amount of contrast in the aorta was not tested.

Figure 7 a) r = 0.220, p = 0.162 b) r = 0.101, p = 0.528

(21)

7 Discussion:

In this research project, CT measurements of changes in the native attenuation, late arterial phase attenuation and of contrast-enhancement in hypervascular SI-NET liver metastases were applied to monitor the effect of ¹⁷⁷Lu-DOTA-octreotate therapy (PRRT). A significant decrease (9 HU  25, CI 1.6 to 17.0) in the metastatic contrast-enhancement was found at CT after end of PRRT as compared to baseline. Also, there was a significant decrease of both the mean (10 HU  23, CI 2.3 to 16.8) and maximum (16 HU  44, 95% CI, 2.6 to 30.1)

attenuation of the metastases in the late arterial phase, but not of their native attenuation.

These findings can, however, not be regarded as clinically significant and represent a basis for therapy monitoring because these attenuation difference could be explained solely by

technical variations between CT examinations as explained later in the strengths and weaknesses section.

Moreover, there was a large variation between patients regarding the changes in metastatic contrast-enhancement from baseline to post therapy CT (range -61 to 50 HU), and regarding the mean and maximum late arterial phase attenuation. Theoretically, these variations during therapy may have reflected differences in response to PRRT between patients. However, a further analysis in which the patients instead were grouped depending on changes in contrast- enhancement and metastatic maximum late arterial phase attenuation (increase >10HU, decrease >10HU and no change) could not establish a relation between these three different radiological therapy responses and the corresponding PFS. The rational was that patients with a large decrease in contrast-enhancement, theoretically reflecting a beneficial effect of the therapy, would achieve a longer PFS and vice versa. However, as stated, no such correlation could be shown. Thus, based on the results of this study, therapy outcome after PRRT in SI- NET patients cannot be predicted by changes in native, arterial-phase attenuation or contrast- enhancement of liver metastases.

7.1 Strengths and weaknesses

An example of technical variations between CT examinations is that the calibration of the CT scanner to water (0HU) can vary between scanners from -3HU to 3HU i.e. up to 6 HU

according to the vendor´s specifications (36). The precision in the attenuation measurements may also change during the day, with less reliable measurements when the scanner is

switched on for the first examination of the day, as compared to later in the day after several examinations allowing for warm-up of the scanner. In most clinical settings, these technical

(22)

variations are of minor concern, but for precise attenuation measurements, these drawbacks need to be taken into account. In a prospective setting, for example when CT is performed for characterization of adrenocortical adenomas, the patients are scanned later in the day when the scanner is warm (37). Because of the retrospective design of the present study the patients were examined different times during the day and these technical factors may therefore have introduced variations in the attenuation measurements with an impact on the interpretation of the present results.

Inherent to the retrospective study design are some other major weaknesses. Because of the strict inclusion criteria, stating presence of both native and late arterial phase at baseline CT as well as post treatment CT, merely a limited number of patients were eligible for evaluation, which could lead to a serious selection bias. A major factor that led to exclusion of patients from the original cohort was the demand for a pre-contrast (native) CT examination at both occasions. Especially for NET patients who survive many years with their disease and

undergo a large number of CT examinations for therapy monitoring, together with three-phase CT of the liver (native, late arterial phase, venous phase), the radiation dose becomes high.

The phase that generally contains the least information is usually the native phase and many standard CT examination protocols therefore exclude the pre-contrast scanning to decrease the radiation dose to the patient but without a high risk of missing important information. Less strict inclusion criteria and a study design concentrating on the arterial attenuation and the contrast-enhancement in the metastases would therefore have yielded a larger cohort with better statistical power and less selection bias. Also, estimation of the progression date is problematic in a retrospective trial, questioning the validity of PFS in a retrospective setting.

All we know is that progression happened between two examinations but as the time interval between two examinations increases, the accuracy of determining the progression date decreases. In the near time after diagnosis and during treatment patients are monitored frequently but as time passes with the disease remaining in a stable state, examinations are being performed less frequently.

Another weakness stems from the fact that the number of PRRT-cycles varies among patients, as mentioned in the introduction. Some of the patients included in this project had been

treated merely with two cycles (n=3) while some received as many as seven cycles (n=5). The majority of patients received four to six cycles. Naturally, this variation itself could explain

(23)

statistical analyses. Also, PRRT requires somatostatin receptor expression in the tumors but as tumors become more malignant they tend to dedifferentiate and lose somatostatin receptor expression. Therefore, the vast majority of PRRT-patients have low grade-tumors (G1-G2) at baseline. Although, in rare cases, perhaps even G3-tumors that for some reason retained their receptor expression, might be eligible for PRRT, contributing to the heterogeneity of the cohort. Baseline tumor grade, which obviously can affect the prognosis of a cancer and the outcome of treatment, was not taken into account in the statistical analysis either. In a study including more patients, these confounders could have been tested for by applying a cox- regression model, which allows for adjusting for various additional categorical covariates in the analysis. Furthermore, the degree of vascularity of the metastases at baseline varied among patients with some metastases being showing very high arterial phase attenuation, while others were slightly more attenuating than the surrounding liver. In a study with more patients or less strict inclusion criteria one might exclude the “not so high attenuating”

metastases and examine only the apparently high attenuating metastases, which in theory reflect areas of the tumor that can be effectively targeted by the treatment, perhaps producing different results.

The contrast enhancement in the aorta at the level of the celiac trunk is an approximation of the contrast-enhancement of the hepatic artery and consequently of the tumor vessels in the metastases. One strength with this study is that in order to exclude that any differences in arterial attenuation and contrast-enhancement in the metastases between CT at baseline and after treatment were merely related to changes in contrast-enhancement technique (iodine concentration, injection rate, contrast volume, time interval between contrast medium injection and scan start), the changes in the arterial attenuation as well as the contrast- enhancement in the metastases were correlated to changes in contrast enhancement in the aorta (Pearson’s product moment correlation). This was, however, found not to be the case with r = 0,101, p = 0,528 and r = 0.220, p = 0.162, respectively rejecting this hypothesis.

Concerning the statistical analyses, the assumptions of a similar proportion and pattern of censoring were not met in the Kaplan-Meier analysis, meaning that the resulting survival curves and median survival times might be incorrect as they could be based on the pattern of censoring rather than on the actual differences between groups. In the Pearson’s correlation coefficient, there were outliers in both tests, that can exhibit a large effect on the coefficient r.

The nonparametric Spearman’s rank correlation coefficient, which is less affected by outliers,

(24)

could have been used instead. However, we decided to go with the more powerful and well- known Pearson’s test anyway since it was clear that there was not a strong correlation after visual inspection of the scatter plots.

7.2 Interpretation of the results

The changes in metastatic contrast-enhancement (T-test 1) and those of the metastatic

attenuation in the late arterial phase (T-test 2) were almost identical between the baseline and follow-up examinations. The rationale behind measuring the contrast-enhancement was that we wanted to make sure that our measurements solely represented changes in contrast-

enhancement and not those attributed to differences in the native metastatic attenuation. Since the native attenuation in a metastasis can change over time for example due to tumor necrosis, anemia or vascular volume overload, comparisons of the metastatic attenuation in the late arterial phase could be influenced by changes in the metastatic attenuation in the native phase.

In our patient cohort there were, however, no differences between the mean change in metastatic contrast-enhancement and the mean change in metastatic attenuation in the late arterial phase between pre- and post-treatment examinations, indicating that none of our patients experienced the conditions that can lead to a change in native attenuation. Therefore, it can be concluded that the results of these two tests provide exactly the same information and both could both be used for further analysis rendering very similar results. We chose to continue with the results from T-test 1.

It is well-recognized that NETs as a group have hypervascular metastases more frequently than other types of tumors but it is difficult to find studies which specify the exact prevalence of hypervascular metastases for the different types of NET. A part of the explanation for the results in this study might therefore be that SI-NET metastases may not generally be as well vascularized as usually described in the literature and not to the same extent as metastases from pancreatic NETs. In support of this is the fact that out of the 233 SI-NET patients initially included in this study at least 48 did not have hypervascular metastases. Probably even many more patients in this cohort lacked hypervascular metastases as some patients were already excluded due to other causes (see Figure 1) before checking whether they had hypervascular metastases.

7.3 Conclusions

There is a need for better radiological methods for evaluating therapy response in

neuroendocrine tumors. Tumor attenuation on CT in the venous phase is used for monitoring

(25)

therapy and predicting outcome in some other types of tumors but has not been studied for SI- NETs until this research project, in which CT measurements of changes in the native

attenuation, late arterial phase attenuation and of contrast-enhancement in hypervascular SI- NET liver metastases were applied to monitor the effect of ¹⁷⁷Lu-DOTA-octreotate therapy (PRRT). There was no change in attenuation or contrast-enhancement in the metastases after treatment that could be regarded as clinically significant. Neither did the small decrease in late arterial phase attenuation, maximum attenuation and contrast-enhancement correlate to clinical outcome measured as progression free survival. Thus, based on this patient selection, it seems that measurements of liver metastasis attenuation or contrast enhancement in SI-NET patients undergoing PRRT are of no use for evaluating the therapy response or for predicting outcome.

The need for better radiological methods for evaluating therapy response in neuroendocrine tumors remains and therefore further studies in this field are needed. A study on SI-NET patients with less strict inclusion criteria might provide more information as to whether attenuation measurements could be useful. It would be even more interesting to conduct a study with a similar protocol but on patients with pancreatic NET because of the indicia that metastases from these NETs are more well-vascularized than those from SI-NETs.

(26)

8 Acknowledgements

Enormt stort tack till Anders Sundin, professor i positronemissionstomografi, för all hjälp och engagemang under skrivandet av detta spännande forskningsarbete!

(27)

9 References:

1. Yao JC, Hassan M, Phan A, Dagohoy C, Leary C, Mares JE, m.fl. One Hundred Years After “Carcinoid”: Epidemiology of and Prognostic Factors for Neuroendocrine Tumors in 35,825 Cases in the United States. J Clin Oncol. 20 juni 2008;26(18):3063–72.

2. Modlin IM, Lye KD, Kidd M. A 5-decade analysis of 13,715 carcinoid tumors. Cancer. 15 februari 2003;97(4):934–59.

3. Sundin A. Imaging of neuroendocrine tumors. Expert Opin Med Diagn. september 2012;6(5):473–83.

4. Hauso O, Gustafsson BI, Kidd M, Waldum HL, Drozdov I, Chan AKC, m.fl.

Neuroendocrine tumor epidemiology: contrasting Norway and North America. Cancer. 15 november 2008;113(10):2655–64.

5. Sundin A. Radiological and nuclear medicine imaging of gastroenteropancreatic

neuroendocrine tumours. Best Pract Res Clin Gastroenterol. 01 december 2012;26(6):803–

18.

6. Scarpa A, Mantovani W, Capelli P, Beghelli S, Boninsegna L, Bettini R, m.fl. Pancreatic endocrine tumors: improved TNM staging and histopathological grading permit a

clinically efficient prognostic stratification of patients. Mod Pathol. juni 2010;23(6):824–

33.

7. Saha S, Hoda S, Godfrey R, Sutherland C, Raybon K. Carcinoid tumors of the

gastrointestinal tract: a 44-year experience. South Med J. december 1989;82(12):1501–5.

8. Marshall JB, Bodnarchuk G. Carcinoid tumors of the gut. Our experience over three decades and review of the literature. J Clin Gastroenterol. mars 1993;16(2):123–9.

9. Sakai D, Murakami M, Kawazoe K, Tsutsumi Y. Ileal carcinoid tumor complicating carcinoid heart disease and secondary retroperitoneal fibrosis. Pathol Int. maj 2000;50(5):404–11.

10. Wang Y-Z, Mayhall G, Anthony LB, Campeau RJ, Boudreaux JP, Woltering EA.

Cervical and upper mediastinal lymph node metastasis from gastrointestinal and pancreatic neuroendocrine tumors: true incidence and management. J Am Coll Surg. juni

2012;214(6):1017–22.

11. Strosberg JR, Nasir A, Hodul P, Kvols L. Biology and Treatment of Metastatic

Gastrointestinal Neuroendocrine Tumors. Gastrointest Cancer Res GCR. 2008;2(3):113–

25.

12. Modlin IM, Kidd M, Latich I, Zikusoka MN, Shapiro MD. Current Status of Gastrointestinal Carcinoids. Gastroenterology. 01 maj 2005;128(6):1717–51.

13. Meijer WG, Veer E van der, Jager PL, Jagt EJ van der, Piers BA, Kema IP, m.fl. Bone Metastases in Carcinoid Tumors: Clinical Features, Imaging Characteristics, and Markers of Bone Metabolism. J Nucl Med. 02 januari 2003;44(2):184–91.

(28)

14. Mallory GW, Fang S, Giannini C, Van Gompel JJ, Parney IF. Brain carcinoid metastases:

outcomes and prognostic factors. J Neurosurg. april 2013;118(4):889–95.

15. Sundin A, Arnold R, Baudin E, Cwikla JB, Eriksson B, Fanti S, m.fl. ENETS Consensus Guidelines for the Standards of Care in Neuroendocrine Tumors: Radiological, Nuclear Medicine and Hybrid Imaging. Neuroendocrinology. 2017;105(3):212–44.

16. Carcinoid metastases to the liver: role of triple-phase helical CT. - PubMed - NCBI [Internet]. [citerad 12 januari 2018]. Tillgänglig vid: https://www-ncbi-nlm-nih-

gov.ezproxy.its.uu.se/pubmed/?term=Carcinoid+metastases+to+the+liver%3A+role+of+tri ple-phase+helical+CT.

17. Essen M van, Sundin A, Krenning EP, Kwekkeboom DJ. Neuroendocrine tumours: the role of imaging for diagnosis and therapy. Nat Rev Endocrinol. februari 2014;10(2):102.

18. Raymond E, Dahan L, Raoul J-L, Bang Y-J, Borbath I, Lombard-Bohas C, m.fl. Sunitinib malate for the treatment of pancreatic neuroendocrine tumors. N Engl J Med. 10 februari 2011;364(6):501–13.

19. Yao JC, Shah MH, Ito T, Bohas CL, Wolin EM, Van Cutsem E, m.fl. Everolimus for advanced pancreatic neuroendocrine tumors. N Engl J Med. 10 februari 2011;364(6):514–

23.

20. Kwekkeboom DJ, de Herder WW, Krenning EP. Somatostatin receptor-targeted radionuclide therapy in patients with gastroenteropancreatic neuroendocrine tumors.

Endocrinol Metab Clin North Am. mars 2011;40(1):173–185, ix.

21. Juhlin C, Calissendorff J, Edekling T, Gimm O, Hellman P, Johanson V, m.fl.

Neuroendokrina buktumörer (GEP-NET) Nationellt vårdprogram [Internet]. 2015 [citerad 10 april 2017]. Tillgänglig vid:

http://www.cancercentrum.se/samverkan/cancerdiagnoser/neuroendokrina- buktumorer/vardprogram/gallande-vardprogram/16.-referenser-och-lankar/

22. Sandström M, Garske-Román U, Granberg D, Johansson S, Widström C, Eriksson B, m.fl. Individualized Dosimetry of Kidney and Bone Marrow in Patients Undergoing 177Lu-DOTA-Octreotate Treatment. J Nucl Med. 01 januari 2013;54(1):33–41.

23. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, m.fl. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer Oxf Engl 1990. januari 2009;45(2):228–47.

24. Sundin A, Rockall A. Therapeutic Monitoring of Gastroenteropancreatic Neuroendocrine Tumors: The Challenges Ahead. Neuroendocrinology. 2012;96(4):261–71.

25. Binderup T, Knigge U, Loft A, Mortensen J, Pfeifer A, Federspiel B, m.fl. Functional Imaging of Neuroendocrine Tumors: A Head-to-Head Comparison of Somatostatin Receptor Scintigraphy, 123I-MIBG Scintigraphy, and 18F-FDG PET. J Nucl Med. 01 maj 2010;51(5):704–12.

(29)

26. Srirajaskanthan R, Kayani I, Quigley AM, Soh J, Caplin ME, Bomanji J. The role of 68Ga-DOTATATE PET in patients with neuroendocrine tumors and negative or equivocal findings on 111In-DTPA-octreotide scintigraphy. J Nucl Med Off Publ Soc Nucl Med. juni 2010;51(6):875–82.

27. Ambrosini V, Tomassetti P, Castellucci P, Campana D, Montini G, Rubello D, m.fl.

Comparison between 68Ga-DOTA-NOC and 18F-DOPA PET for the detection of gastro- entero-pancreatic and lung neuro-endocrine tumours. Eur J Nucl Med Mol Imaging.

augusti 2008;35(8):1431–8.

28. Benjamin RS, Choi H, Macapinlac HA, Burgess MA, Patel SR, Chen LL, m.fl. We

Should Desist Using RECIST, at Least in GIST. J Clin Oncol. 01 maj 2007;25(13):1760–4.

29. Choi H, Charnsangavej C, Faria SC, Macapinlac HA, Burgess MA, Patel SR, m.fl.

Correlation of Computed Tomography and Positron Emission Tomography in Patients With Metastatic Gastrointestinal Stromal Tumor Treated at a Single Institution With Imatinib Mesylate: Proposal of New Computed Tomography Response Criteria. J Clin Oncol. maj 2007;25(13):1753–9.

30. Jiang T, Kambadakone A, Kulkarni NM, Zhu AX, Sahani DV. Monitoring Response to Antiangiogenic Treatment and Predicting Outcomes in Advanced Hepatocellular

Carcinoma Using Image Biomarkers, CT Perfusion, Tumor Density, and Tumor Size (RECIST): Invest Radiol. januari 2012;47(1):11–7.

31. Vyas SJ, Puri YS, John BJ, Yu D, Watkins J, Imber C, m.fl. Radiological tumor density and lymph node size correlate with survival in resectable adenocarcinoma of the pancreatic head: A retrospective cohort study. J Cancer Res Ther. mars 2016;12(1):417–21.

32. Smith AD, Shah SN, Rini BI, Lieber ML, Remer EM. Morphology, Attenuation, Size, and Structure (MASS) criteria: assessing response and predicting clinical outcome in

metastatic renal cell carcinoma on antiangiogenic targeted therapy. AJR Am J Roentgenol.

juni 2010;194(6):1470–8.

33. O’Riordan E, Craven CM, Wilson D, Robinson PJ. Dual Phase Hepatic CT. Am J Roentgenol. 01 maj 2000;174(5):1417–21.

34. Gutman SI, Piper M, Grant MD, Basch E, Oliansky DM, Aronson N. Progression-Free Survival: What Does It Mean for Psychological Well-Being or Quality of Life? [Internet].

Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 [citerad 20 november 2017]. (AHRQ Methods for Effective Health Care). Tillgänglig vid:

http://www.ncbi.nlm.nih.gov/books/NBK137759/

35. Goldman LW. Principles of CT and CT Technology. J Nucl Med Technol. 09 januari 2007;35(3):115–28.

36. Sundin A. Imaging of Adrenal Masses with Emphasis on Adrenocortical Tumors.

Theranostics. 17 maj 2012;2(5):516–22.

Therapy monitoring of small intestinal neuroendocrine tumors with CT; quantification of tumor attenuation and contrast-enhancement