A. Prueba Kolmogorov – Smirnov
2.9 PROBABILIDAD, PERIODO DE RETORNO Y RIESGO DE FALLA
tumour free.
This section will consider the limitations o f currently available imaging techniques.
5 2 . 4 . 1 Ra d i o n u c l i d e Im a g i n g
This technique has been the mainstay of liver tumour detection for many decades. It is easy to perform and relatively inexpensive. The smallest diameter lesions detectable is in the order o f 1.5 to 2.0 cm using the most recent technical advances. Despite the inherent difficulties in confirming the presence or absence o f liver involvem ent as previously discussed, sensitivities o f 70 to 80% are quoted for detecting liver lesions. However, the main problem is that all space occupying lesions within the liver appear the same on radionuclide imaging. The low specificity (Lima et al., 1975, Ostfeld & M eyer., 1981) inevitably requires a second investigation to determine the nature o f the lesion.
Compared to CT and U S, the lack o f spatial resolution in particular to major vascular and biliary structures coupled with poor tissue characterization are major drawbacks. These factors limit the adoption of liver scintigraphy as a primary diagnostic tool in the detection o f liver cancer.
5 2 . 4 . 2 He p a t i c f l o w Sc i n t i g r a p h y
This technique has been described as a method o f enhancing sensitivity and specificity o f hepatic imaging. Unlike the normal liver which receives 20% o f its total blood supply from the hepatic artery, métastasés derive almost their entire blood supply from this source. This inturn produces an altered flow pattern with the liver receiving an increased proportion o f arterial blood. This can be quantified using a scintigraphic analysis expressing the ratio o f arterial blood flow to total hepatic blood flow.
In a prospective study, 151 patients were studied with the absolute determinant of liver m étastasés being palpation and follow up (L evenson et al., 1985). A reported sensitivity o f 96% and a specificity o f 72% were reported. However, lesions under 2 cm in diameter cannot be imaged by radionuclide techniques and therefore this technique is unlikely to have a large role in screening. Its second drawback is the lack o f information regarding relations o f métastasés to major vascular structures.
5 2 . 4 . 3 Ul t r a s o u n d ( U S )
This technique has many factors in its favour. It is w idely available, non-invasive, simple to perform with rapid examining speeds. Its main weakness is that its efficacy is highly operator dependant with enthusiasts reporting excellent results. It is generally accepted that focal hepatic lesions less than 2 cm in diameter cannot be reliably detected by US. The most optimistic study using linear array transducers report detection o f lesions as small as 11 mm (Sheu et al., 1985). There is no strong correlation between sonographic appearance and histological type (Green et al., 1977), therefore, tissue specificity is poor making fine needle aspiration and core biopsy o f suspected métastasés a necessity for confirmation. When combined, such an approach can achieve sensitivity and specificity rates o f 92% and 100% respectively making it as accurate as laparoscopic guided liver biopsy but much less invasive (Montali et al., 1982). In a prospective study comparing U S, CT and scintigraphy to detect biopsy proven m etastatic cancer, there was no significant differences in sensitivity and specificity between these three imaging modality (smith et al., 1982).
A recent useful extension of US is its intra-operative application for the detection of small (diameter < 2cm) intrahepatic focal lesions. The absence o f a soft tissue interface between the transducer and the liver allows higher frequencies (5 to 12.5 MHz) to be used enhancing spatial resolution. This approach is not in w ide use but it is gaining increasing recognition. Thomas et al (1987) undertook an in-vivo study in 20 post mortem livers removed from patients who died from gastric or colorectal cancer and determined the correlation between direct contact US and histological examination o f serial slices o f liver for detecting liver métastasés. O f 67 lesions, 45 were successfully imaged (67%), 35 out o f 37 lesions greater than 1 cm (95%) were detected, while 10 out of 15 métastasés 0.5 to 1.0 cm were imaged. None of the 15 métastasés less than 0.4 cm in diameter were picked up. The authors had no previous experience yet there were no false positives.
An evaluation of intra-operative ultrasound (lO U S) was carried out by Machi and his colleagues (1987) in 84 patients undergoing surgery for colorectal cancer. A ll patients had preoperative US and CT scans o f the liver. O f 46 métastasés detected, 32 were imaged by preoperative investigation and or palpation o f the liver. O f these, 31 were detected by lOUS in addition to a further 14 giving an overall detection rate o f 98% (45/46) with an increased detection rate of 28% over preoperative assessment. All métastasés were less than 2 cm in diameter and located deep within the liver substance. In a similar study, Kane and his colleagues (1987) found a 25% increased detection rate, the higher sensitivity rate favouring small lesions. In none o f these studies is a false positive rate quoted. Several other workers have confirmed the increased sensitivity o f lOUS and in addition, assessed the impact o f this technique on surgical decision making ( Gozzetti et al., 1986, Machi et al., 1986, Rifkin et al., 1987, Bismuth et al., 1987). AU reported that lOUS provided additional information in up to 30% o f all patients resulting in a change of management or surgical approach.
lOUS reliably detects métastasés down to 0.5 cm in diameter deep within the liver substance with its associated im plications on management. What influence this has on patient survival is not clearly known and is likely to be difficult to assess in a randomised controlled trial as a non-IOUS arm maybe unacceptable on ethical grounds. However, its ability to delineate the relationship between tumours and intrahepatic biliary and vascular
structures will ensure it has an important role in the planning o f hepatic resections. The routine adoption o f lOUS at the time o f primary tumour resection would be an ideal goal for accurate tumour staging. H owever, it may be impractical in those instances where exposure o f the liver involves extending the incision with prolongation o f operating time.
5 2 . 4 . 4 COMPUTERISED TOMOGRAPHY (CT)
This is the commonest method employed for detecting hepatic lesion and depends on differences in the x-ray attenuation pattern between métastasés and normal liver parenchyma. This, in practice, is very little and various water soluble contrast agents are employed to enhance the attenuation o f the normal parenchyma or the metastasis.
In a review, Clark and Matsui (1983) defined 7 methods o f CT scanning with water soluble contrast agents. The contrast may be administered intravenously in one o f two ways. The first is by intravenous infusion or alternatively by rapid bolus administration with immediate dynamic sequential scanning as the contrast is injected. It is possible to combine both these methods which rely on higher uptake o f contrast by the normal parenchyma than the tumour. The bolus technique however achieves a better differential as there is less time for the contrast to diffuse into the tumour. However, this method requires third or fourth generation scanners capable o f rapid scan times with shon interscan delays. A disadvantage of this approach is that large doses o f contrast may need to be given leading to nausea and vomiting in many patients.
An alternative approach is to administer the contrast intra-arterially (CT arteriography). Hepatic métastasés derive their entire blood supply from the hepatic artery unlike the normal liver which only derives 20% o f its total supply from this source. Injection o f contrast into the hepatic artery exploits this fact leading to increased opacification o f the métastasés. CT arterial portography relies on injection o f the contrast into the superior mesenteric artery producing selective opacification o f the portal venous blood. Unlike métastasés, the normal parenchyma shows increased opacification. CT arteriography and portography are invasive and costly. In addition, the necessity for hospital admission and morbidity o f an angiogram make both these techniques unsuitable for routine use.
Delayed scanning provides another way o f delineating between normal parenchyma and métastasés. Three to six hours follow ing contrast administration, there is increased opacification o f normal liver due to contrast in the hepatocytes and biliary systems. Métastasés, unable retain the contrast for this length o f time appear as filling defects. The disadvantages o f this approach include large contrast doses, doubling o f the radiation doses and examination times with their implication on cost.
EOE-13 is a 53% aqueous em ulsion o f iodinated poppy seed oil developed by V em ies and his colleagues at the National Institute of Health (1979). This material in taken up selectively by reticulo-endothelial cells in the liver and spleen but not by métastasés within them thereby allowing selective opacification o f normal parenchyma. However, EOE-13 is relatively unstable with limited availability and in up to 60% o f patients produces chills and fever (Miller et al, 1984). It is regarded by many as an experimental agent only.
There is now a wealth o f evidence recognizing that contrast enhancement is essential for reliable detection o f liver métastasés, however, much controversy surrounds the optimal method o f delivery and the timing o f the scans. In a prospective study by Pausher et al (1989), hepatic CT scans were performed in 50 patients using no contrast compared to rapid contrast infusion and bolus contrast administration with rapid sequential scanning. The latter technique detected 10% and 15% more lesions than the infusion approach and non-contrast imaging respectively. Other workers have also shown the superiority o f bolus administration detecting as many as 40% more lesions than non contrast and infusion approaches (Berland et al., 1982, Burgener & Hamlin., 1983). Using bolus administration, Freeney et al (1986) showed a 73% sensitivity and a 99% specificity
I
in detecting liver métastasés from colorectal cancer confirmed by histology at laparotomy. Bernardino and his colleagues (1986) attempted to assess the value o f delayed scans combined with bolus administration and dynamic sequential scanning in the same patient. Both these techniques identified the same number o f patients, the delayed scan providing better definition o f metastasis architecture while detecting more lesions in 27% o f patients. However, som e 11% o f lesions were better visualised on the dynamic images. There were no false positives with either technique and on balance the two techniques would seem complimentary.
Other investigators have attempted to improve the accuracy o f CT by injecting contrast material into the superior mesenteric artery (CT portography) or the hepatic artery (CT arteriography). Matsui and his colleagu es (1987) compared the sensitivity o f conventional ultrasound, dynam ic sequential CT after bolus administration, CT arteriography, CT portography and CT after intra-arterial injection o f iodinated poppy seed oil in detecting 45 colorectal hepatic métastasés resected in 22 patients. The respective sensitivities were 58%, 63%, 50%, 84% and 38%. O f 18 lesions less than 15 mm in diameter, 10 were detected by CT portography alone. Five out o f 7 lesions not visualised by CT portography were less 5 mm in diameter, however, this technique was falsely positive in 4 out of the 22 patients. The low relative specificity o f CT portography has also been reported by other workers (Miller et al., 1987) making ultrasound or conventional CT essential to exclude benign lesions such as cavernous haemangioma, cysts, adenomas, fibronodular hyperplasia and abscesses. W hile invasive and specialized, CT portography is useful in delineating the relationship o f métastasés to major vascular structures. The low detection rate reported by Matsui and his colleagues for EOE-13 has not been confirmed by other workers. For example. M iller et al (1984, 1987) in a comparison o f EOE-13 enhanced scans and surgical histological findings noted a 77% detection rate which was twice as many as either unenhanced or water soluble contrast enhanced CT.
The difficulty o f comparing sensitivity and specificity o f imaging techniques in different series has been previously discussed. With CT imaging, this problem is further compounded by non-unifomi methods (contrast volume and administration rate, timing and types o f CT scanners) practiced for any given enhancement technique. Thus only broad generalisations can be made regarding the optimal CT enhancement method.
As discussed in the next chapter, CT scanning formed the principle im aging modality in detecting and assessing the response o f liver métastasés following interstitial laser therapy. Tumour detection at the earliest possible stage is likely to determine the results o f ILH and subsequent benefit to patients. It is therefore important that the optimum contrast technique for CT imaging o f métastasés is employed. On the basis o f the literature review presented in this section, the combined bolus and infusion technique with delayed
scans probably represents the optimum and most practical method o f CT im aging o f focal hepatic lesions.
All im ages presented in chapter 6 were obtained using a Siem ens CT scanner (Somatom, DR). Detection of métastasés was performed using contiguous 8 mm slices. Oral and intravenous contrast were given. The latter as a bolus and an infusion during scanning. Oral contrast was administered 20 minutes and immediately before initial liver scanning. Scans through the same area were repeated immediately follow ing 100 ml bolus o f intravenous contrast plus 50 ml during scanning. Delayed scans through the same area were repeated 45 to 60 minutes after injection. Post laser im aging consisted o f liver scanning immediately following bolus injection o f 100ml of intravenous contrast with 50 ml administered during scanning.
5 2 . 4 . 5 Ma g n e t i c r e s o n a n c e Im a g i n g ( M R I )
MRI is an effective method for detecting hepatic métastasés which does not require ionizing radiation or injection o f toxic or allergenic substances. It is not the place o f this review to discuss the detailed mechanics o f this imaging modality but as with CT scanning, results vary with technique and instrumentation.
The real question posed by MRI is whether it has any useful advantages compared to CT scanning ? In a prospective multi-institutional study, Chezmar et al (1988) compared the detection rate o f hepatic métastasés in 59 patients using CT (Dynam ic sequential -f delayed scans) versus MRI. The absolute determinant for the presence o f hepatic métastasés were surgical exploration, biopsy and follow up. Of 28 patients with malignant liver disease, MRI demonstrated this in 27 (96% sensitivity) and CT in 26 (93% sensitivity). However, the false positive rate for MRI was 26% (8/31) compared to 6% (2/31) for CT. O f 16 patients with significant extrahepatic findings, CT detected this in 12 and MRI only in 1. This difference is statistically significant and is confirmed in similar studies by Glazer et al (1986) and Stark et al (1987).
H owever, differences between CT and MRI seem not to be clear cut. Several studies have focused on comparing the efficacy o f CT and MRI for detecting focal hepatic lesions. Som e have found CT superior (Glazer et al., 1986, N elson et al., 1988) while
others have found the converse to be the case (Reinig et al., 1987, Stark et al., 1987). These differences probably reflects various CT and MR techniques used with different equipment and design studies. The drawbacks o f MRI are limited availability, prolonged examination times and limited depiction o f extrahepatic structures. For the moment, the real differences between CT and MRI are probably small. The availability, experience and superior extrahepatic delineation would favour CT especially when used in dynamic sequential mode after bolus administration (an accurate indicator o f patients with hepatic disease) with delayed imaging (an accurate indicator of the number o f lesions).
MRI perform ance may be im proved with the use o f contrast m edia. A supraparamagnetic iron oxide agent which is taken up selectively by normal liver tissue has been tried with worthwhile improvements (Stark et al., 1988). For the moment this agent must be regarded as experimental.
5.3 TREATM ENT
The value o f any treatment is judged by its influence on the natural history o f the disease process. It may prolong life and or improve its quality. The trade o ff for the patient for such gains must be an acceptable low risk o f serious com plications. Given the retrospective nature o f most studies, the ethical implications o f withholding any treatment however experimental for a condition which is uniformly fatal, it is debatable whether the natural history o f liver cancer is fully understood. Therein lies the conundrum. H ow to assess the curative or palliative value o f any treatment when formidable ethical and moral obstacles prevent a scientific approach to understanding the natural history o f a common problem. However, as clinicians we have to make a choice between therapeutic nihilism and unjustified aggressiveness involving substantial risk to the patient.
5.3.1 HEPATIC RESECTION
The first hepatic resection for colorectal métastasés was performed over 100 years ago in Germany (Garre., 1888). In the last 10 years, there has been a large increase in the number o f minor and major hepatic resections carried out due in part to a greater
understanding o f the segmental anatomy o f the liver, improved surgical and anaesthetic techniques and a more aggressive approach to the management o f hepatic tumours.
Hepatic resection is the only treatment which offers the prospect o f cure. However, it is only appropriate for a small group o f patients variously estimated at 5 to 10% of those with colorectal métastasés and 9 to 17% with hepatocellular carcinoma (Fortner et al., 1984, Okuda., 1980, Lee et al., 1982). Much controversy surrounds the true value o f hepatic resection. In this section I shall put forward both sides o f the argument. Due to their high frequency, colorectal hepatic métastasés as a group have been best studied and provide the hard core o f available data. In addition, their biological behaviour is more favourable than upper gastrointestinal cancers and it is relatively easy to achieve locoregional control of the primary tumour while the pattern o f haematogenous spread is relatively well defined.