When not to treat (random thoughts)
It's a really, really tough question.
We surgeons face this question as well, although probably much less frequently than medical oncologists do (another reason I don't think I could do medical oncology). One of the more common oncologic problems that force us to consider such choices is malignant bowel obstruction in a patient with a known metastic solid tumor. Do we operate to try to relieve the obstruction, even knowing that bowel obstruction in a patient with a metastatic solid tumor is usually a pre-terminal event? It's usually a fairly easy decision if the patient has widespread metastatic disease in multiple sites and isn't likely to live more than a few weeks. On the other hand, if disease in the abdomen is the only known site of disease, it can often take a somewhat indolent course, meaning that the patient might live as long as a few months. In fact, the mean survival for a patient operated on for malignant bowel obstruction is around five or six months. If no operation is done, the patient will spend those remaining months never being able to eat again. He will have to live out his days on intravenous nutrition, with either a tube sticking out of his side (called a gastrostomy tube) to drain the backed-up intestinal fluid from his stomach and thus keep him from constantly vomiting or (worse) a tube in his nose to do the same thing (which sometimes happens in the case of patients with such widespread intraabdominal disease that using endoscopy to put a gastrostomy tube in is not safe. Some patients in this situation will opt to forgo intravenous fluids and nutrition, in which case they will die of dehydration within hours to days.
Of course, the question then becomes: Are the potential benefits of surgery worth the risks in these patients? As is the case for chemotherapy, the risks of surgery are substantial. Overall, explorations related to malignant bowel obstruction carry around a 10% mortality rate and a 30% rate of major complications. In older, sicker patients, these risks will be higher, sometimes much higher. Worse, the rate of success is relatively low. In the case of obstructions from benign causes (adhesions from previous surgery, for instance), success, as measured by the chance of the patient's leaving the hospital able maintain his hydration and nutrition by oral intake alone, will be very high. In the case of a localized tumor obstruction (an obstructing rectal cancer, for instance, with no disease detectable elsewhere in the abdomen), success will still be pretty high, but not nearly as high as for a benign obstruction. In the case of more extensive intraabdominal disease (called carcinomatosis), the chances of getting the patient out of the hospital able to eat and drink enough to survive are fairly low. Worse, if there are significant complications (not uncommon), the patient may linger in the hospital, spending his brief remaining time away from home and family, suffering numerous pains and indignities, and ultimately dying in the hospital or (if he's lucky) getting sent to a good hospice.
Given these realities, a less interventionalist approach is usually better. Certainly, any patient with a previously treated solid tumor who develops a bowel obstruction should be considered for surgery. If no cancer is seen in the abdomen on imaging studies, there is still a substantial chance that this might be a benign obstruction, particularly if the patient has undergone prior surgery and therefore might have adhesions. In general, these patients should be given the benefit of the doubt and operated on, unless they are poor surgical risks. Patients with known intraabdominal disease who develop an obstruction should probably be treated with more circumspection. For good operative candidates, a good general rule is: "Everybody deserves one chance." If the abdomen is the only known site of disease, a tendency is err on the side of operating to see if anything can be done (unless the patient is a bad surgical risk) is a reasonable policy, because it really sucks not to be able to eat and to have to have a tube sticking out of one's stomach or nose for the rest of one's life. (Giving the patient the ability to eat for a few months is a wonderful thing if it can be accomplished with low enough risk; it should not be underestimated in its effect on quality of life.) However, if the patient has widely metastatic disease to multiple organs, it's usually better to put in a gastrostomy tube, because these patients are not likely to live very long no matter what is done. Finally, for patients with advanced disease and a recurrent bowel obstruction, reoperation is seldom indicated (IMHO). The success rate is just too low for recurrent obstruction to justify it, except in very select cases (very healthy patients, low volume disease on imaging studies, for instance--quite uncommon).
Of course, there are surgeons who believe that huge operations to debulk intraabdominal solid tumors, followed by intraabdominal chemotherapy, will prolong survival. They will even publish and present what appear to be impressive results. To me, it's still unclear whether this is simply selection bias, in which the patients they choose to operate on tend to be healthier people with less extensive disease, who are more likely to live longer anyway. The only exception is ovarian cancer, for which there is good evidence that debulking improves chemotherapy responsiveness. I'm tend to be of the opinion that, as of now, such procedures should not be done except in the context of a clinical trial.
In all cases, however, one must keep in mind the principle of always considering foremost what the patient wants. For this, Dr. Hildreth's four principles are an excellent set of guidelines to keep in mind. In the case of surgery, I would add an additional principle: Even if the surgery goes perfectly, consider how long the patient will be in the hospital versus his life expectancy. If the operation is likely to keep the patient in the hospital two weeks if things go perfectly (and a month, or even more, if they don't) but he is likely only to have two or three months more to live, is it worth it to operate? In this case, as always, it's probably a good idea to defer to the patient's desires when possible.