Monday, April 29, 2019

Is there an oligometastatic state for prostate cancer?

The concept of an "oligometastatic state" is that there exists an early stage where metastases are few in number and are in some way different from metastases that develop later. It also means that there are no micrometastases in systemic circulation (in bone and lymph) and in reservoirs like bone, nerve cells, lymph nodes and other organs. If such a state exists, the cancer can be picked off, like dandelions in a lawn, and the person can be cured.

The alternative concept is that cancer spread is always polymetastatic. Thousands of cells are released from the primary tumor. They find their way to sites where they change the tissue they land in, making it amenable to future growth. This is called "seed and soil." A metaphor might be mushrooms growing at the base of an oak tree. The mycelium extends everywhere throughout the soil and into the roots of the tree. Occasionally, a mushroom crops up. You can pick all the mushrooms you want, but the fungus is never destroyed. There is no way to destroy the fungus short of destroying the roots of the oak tree and sterilizing the soil. This is what "systemic" means.

It is well known that tumor cells must undergo a genomic change called epithelial-to-mesenchymal transition (EMT) before they are capable of traveling and living outside of their original environment. Metastasized cells do not look like or behave like the original tumor in its original tissue; they are phenotypically different.

Are all cancers alike?

There are certain "hallmarks of cancer." To qualify as a cancer, it must be malignant, destroying healthy tissue. Most cancers multiply rapidly, losing the ability to self-destruct when its DNA goes awry (apoptosis). They are usually immortal and evade destruction by the immune system. They can travel from one place to another. Solid tumors change the structure of their host tissue and usually generate their own blood supply and nerve innervation (see cancer as a tissue-based disease).

But all cancers are different. Unlike most other solid tumors, prostate cancer is usually originally multifocal in the prostate. While some cancers can be cured by surgically removing the original tumor, the whole organ must be removed (or irradiated) for prostate cancer. Foci may be a centimeter or more apart, so it is known to have a strong signalling mechanism that changes host tissue. It has a predilection for lymph nodes and bone, where it usually creates osteoblastic lesions (bone overgrowth). It is activated by an androgen receptor, which eventually becomes impervious to androgen deprivation. Tumors tend to be hypoxic, and have low immune-cell infiltration. They are relatively radioresistant, and are not appreciably killed off by non-taxane chemotherapy. There are multiple growth pathways - block one and others predominate. It is also abnormally slow growing. It may take many years for EMT cells to originate. The time from the first detectable metastasis to the second may be years apart. Unlike other cancers, prostate cancer metastatic cells generate energy for reproduction from lipid metabolism at first. Many years later, glycolysis may come to predominate (as it does in most other cancers).

To determine if there is such a thing as an "oligometastatic state" it is therefore necessary to show that such a state exists for every kind of cancer. The first step is to show plausibility. With high throughput sequencing it may be able to distinguish the genomics of early metastases from later ones. However, because genetic breakdown is a characteristic of cancer, it is also necessary to show that the early clones are phenotypically different from later clones. If early clones lack the ability to disseminate and prepare the "soil" for metastatic progression, that would create a case for an oligometastatic state.

It is also necessary to show that such a state exists for every type of cancer, or at least to find the cancers in which such a state exists. One cannot just assume that all cancers are alike in this regard.

Iyengar et al. reported the results of a small trial where 14 patients with non-small cell lung cancer (NSCLC) with a small number of distant metastases were treated with SBRT to the lung and distant lesions ("consolidative therapy") and chemotherapy. 15 patients only received chemo. There was a significant increase in progression-free survival: 9.7 months vs 3.5 months. Larger trials (NRG LU002 and SARON) will investigate overall survival.

(update 12/6/23) NRG LU002 will not move to a Phase III trial because the Phase II Progression-Free Survival failed to meet its prespecified goal.

(update 6.2.22) Steven Chmura reported at the 2022 ASCO meeting (J Clin Oncol 40, 2022 (suppl 16; abstr 1007) that the randomized trial (NRG BR-002) of SBRT (+ standard of care (SOC)) to oligometastatic breast cancer failed to slow progression or increase survival over SOC alone. There were 125 patients (65 SBRT+SOC, 60 SOC). An expanded trial was canceled due to futility.  Clearly then, there is no oligometastatic "state" that applies to all cancers.

The SABR-COMET Phase 2 Trial

Palma et al. recruited 99 patients at 10 hospitals in Canada, Scotland, Australia and the Netherlands from 2012-2015. Patients had 1-5 metastases, and were randomly assigned to high-intensity metastasis-directed radiotherapy (SABR or SBRT) or systemic standard of care. After 2 years median follow-up, there were:

  • 66  patients in the SABR group
  • 33 patients in the control group
  • Most had 1-3 metastases: 94% in the control group, 93% in the SABR group
  • SABR dose was most commonly 35 Gy in 5 treatments,  60 Gy in  8 treatments, and 54 Gy in 3 treatments
  • 12% received additional SABR for disease progression


After a median follow-up of 25-26 months:

  • Overall mortality was 36% for SABR, 48% for control (Hazard Ratio = .75)
  • Overall survival (median) was 41 months for SABR, 29 months for control (Hazard Ratio = 0.57; p=0.09) Note: they prespecified that anything above 80% confidence would be sufficient to expand to a Phase 3 study.
  • 39% had metastatic progression in the SABR group, mostly new metastases
  • 61% had new metastases in the control group
  • Grade ≥2 adverse events: 9% in the control group, 29% in the SABR group
  • 5% of the SABR group died as a result of treatment: radiation pneumonitis, pulmonary abscess, and subdural hemorrhage from surgery to repair a perforated gastric ulcer


The authors are cautious about the toxicity, but optimistic that their study provides proof of an "oligometastatic state." They have already announced two Phase 3 randomized clinical trials for people with 1-3 metastases and 4-10 metastases.

Skewed Distribution of Cancers Accounts for the Purported Benefit

The distribution of cancer types was vastly different in the SABR and control groups. Metastatic colorectal cancer, which has an 70% 2-year mortality rate, is twice as likely to appear in the control group as the SABR group; while metastatic prostate cancer, which has a 10% 2-year mortality rate is more than 3 times as prevalent in the SABR group. This skewed distribution accounts for almost all of the difference that the authors attribute to a treatment effect.


Type of CancerControlSABR
expected 2-year survival (approx)
n% of totaln% of total
Breast515%1320%50% (1)
Colorectal927%914%30% (2)
Lung618%1218%10% (3)
Prostate26%1421%90% (4)
Other1133%1827%
TOTAL3366
(1) https://www.nature.com/articles/bjc2015127
(2) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2739317/
(3) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3096514/
(4) https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)32486-3/fulltext


ControlSABR
Expected 2-year survival due to distribution35%48%
Expected 2-year mortality due to distribution65%52%
Hazard Ratio due to skewed distribution80% 
Reported Hazard Ratio75%

Because of the uneven distributions in the treatment and control, the hazard ratio does not reach 80% confidence and the hypothesis should be rejected. The authors believe that just eliminating prostate cancer patients from both groups would correct the flaw, but they would have to eliminate colorectal cancer as well - I doubt the result would be significant with even 80% confidence. I believe the authors of the study erred in accepting the results even with 80% confidence for forging ahead with a Phase 3 randomized trial. The treatment effect, if any, is so small that their Phase 3 trial as specified is insufficiently powered to detect a treatment effect. They do not propose to stratify by type of cancer. Also, much longer follow-up is needed for prostate cancer.

On top of that, they have not made the case for an oligometastatic state, which would have to be true for every cancer type and not just a weighted average sum of them. They would also have to include genomic and phenotypic analysis of biopsied tissue when there are both few metastases and many in order to demonstrate plausibility.

Patients should note the mortality rate attributable to SABR of metastases. There is little risk in irradiating metastases occurring in safe locations, like the pelvic bones. There may be unacceptable risk in irradiating metastases near the heart, lungs, or digestive tract. Since there is no evidence that metastasis-directed therapy for prostate cancer improves survival, patients should not avoid systemic therapy (for which there is convincing evidence). Patients who are interested in SABR of metastases should talk to experienced radiation oncologists in large tertiary-care facilities.

(update 05/25/2022) SABR-COMET was updated with 8-year results. They report:

  • 8-yr overall survival was 26.2% in the SABR arm vs 13.2% in the control arm (HR=0.50)
  • 8-yr progression-free survival was 21.6% in the SABR arm vs 0.0% in the control arm (HR=0.45)
  • Rates of acute or late Grade 2+ toxicity were 30.3% in the SABR arm vs 9.1% in the control arm

The Kaplan-Meier survival curves showed a large drop-out of participants after 4 years in the control group. This is consistent with the relative lack of prostate cancer patients and the relatively large presence of colorectal cancer patients in the control group. 

There was no difference between arms in time to new metastases. So, the larger progression-free survival in the SABR arm is entirely due to local control. The cancer continued to seed new metastases at the same rate in both arms.



Sunday, April 28, 2019

If you are using PSA to monitor your cancer, you may want to avoid curcumin (and some other supplements)

Curcumin (a turmeric extract) is one of the most popular natural substances subjected to pre-clinical research. Based on mouse and lab studies, it has been touted as the cure to cancer and just about everything else, with reports of activity including anti-inflammatory, anti-HIV, antibacterial, antifungal, nematocidal, antiparasitic, antimutagenic, antidiabetic, antifibrinogenic, radioprotective, wound healing, lipid lowering, antispasmodic, antioxidant, immunomodulating, anticarcinogenic, and Alzheimer’s disease, among others. This "panacea" and the low level evidence behind it are satirized in this amusing video.

It is one of the most widely researched supplements - in mouse and lab studies. In spite of its spectacular success with mice, randomized clinical trials in humans have been lacking. Choi et al. reported on a double-blinded randomized clinical trial of curcumin on 82 evaluable men who completed one treatment cycle of intermittent hormone therapy. They were then given 1440 mg/day of curcumin or a placebo for 6 months. The goal of the study was to see whether curcumin could extend their time off of hormone treatment.
  • Those taking curcumin were able to avoid hormone therapy for 16.3 months
  • Those taking the placebo were able to avoid hormone therapy for 18.5 months
  • The difference was not statistically significant
  • 10% of patients taking curcumin had PSA progression during the curcumin treatment period vs 30% of those taking the placebo.
The fact that those taking the placebo had an insignificantly longer break from hormone therapy in spite of the fact that their PSA progression was greater than those who were taking curcumin in the first 6 months, indicates that curcumin may have interfered with the PSA tests while they were taking it. Clearly, curcumin did not delay clinical progression.

Ide et al. found in a small (n=85) double-blind randomized clinical trial that a mixture of soy isoflavones and curcumin suppressed the serum PSA readings of men with high PSA (>10 ng/ml) who were confirmed by biopsy to not have prostate cancer. The curcumin mixture suppresses the PSA reading independent of prostate cancer.

A multi-center, blinded randomized clinical trial in France of curcumin+docetaxel compared to docetaxel alone in men who were metastatic and castration-resistant was ended early because of futility. Although not statistically significant, combining docetaxel and curcumin consistently gave worse outcomes (progression, survival) than docetaxel plus placebo.

Fabiani et al. reported on 50 consecutive patients with PSA over 4.0 ng/ml or PSA velocity > .75 ng/ml/year. They were given curcumin for 30 days.
  • Baseline % free PSA was 17%
  • After 30 days of curcumin, % free PSA was 20%
  • The changes in PSA and % free PSA were statistically significant
It seems that curcumin suppressed PSA. Although it is possible that 30 days of curcumin reversed the prostate cancer, that is unlikely. It is more plausible that curcumin affected the PSA assay.

This effect has been noted in the literature. The authors of this analysis and this one label curcumin as a Pan-Assay Interference Compound (PAINS), which means that it is known to interfere with assay readouts. Curcumin particularly confounds tests of molecules, like prostate specific antigen (PSA) and prostate-specific membrane antigen (PSMA), that penetrate the cell wall. According to this analysis, other common supplements that may interfere with the integrity of the cell wall without actually binding to a site on the proteins (which would be a real drug effect) include genistein (a soy isoflavone), EGCG (green tea), resveratrol (grapes),  and capsaicin (chili peppers).  Some of these compounds, including curcumin, are capable of forming stable metal ion complexes and should be scrupulously avoided by patients taking Ga-68-PSMA-11, Lu-177-PSMA-617, technetium bone scan, or gadolinium MRI contrast agent. Glutamate (MSG used in Chinese food) is also a powerful chelator. It has been found to markedly decrease the effectiveness of PSMA theranostics (see this link).

There are other supplements that may mask PSA readings without affecting progression. These include saw palmetto, pygeum, and beta-sitosterol. 5-alpha-reductase inhibitors (Proscar and Avodart) affect PSA in men with BPH and prevent the occurrence of prostate cancer. Because they affect PSA in a known way in men with BPH, we are able to correct for the PSA aberration (by doubling the PSA reading). The FDA has warned that biotin, in many multivitamin preparations, may interfere with many laboratory blood tests. Men taking statins should also be aware that it may produce artificially low PSA readings (see this link and this one). Statins, which seem to be beneficial in some observational studies but not in others, may only have an apparent benefit because of masking of PSA, as in this study and this study.

In designing future clinical trials on curcumin, like this one or this one that tests its benefit as an adjuvant therapy to active surveillance, it is important that the measured endpoint not be dependent on PSA. PSA doubling time, biochemical recurrence-free survival, and time before ADT is initiated (which is usually given as a result of increasing PSA) are artificially increased by curcumin. Only endpoints like radiographic progression-free survival and metastasis-free survival are useful. Incidentally, this is also why those endpoints must be chosen when evaluating the effectiveness of metastasis-directed therapy, which will lower PSA arising out of macroscopic metastases but may or may not slow the cancer's progression.

I spent a very short career as a chemist developing radioimmunoassays for biological substances, like PSA, that were only detected in serum in nanomolar and picomolar amounts. I can attest that even small amounts of impurities that adsorb, quench fluorescence, or react with the protein or its antibody can completely invalidate a test. Curcumin seems to do this.

The biggest problem with curcumin as a medication is its oral bioavailability, which is less than 1% and its elimination half-life, which is about a half hour in rats. It is doubtful that enough is bioavailable to have any therapeutic effect. This is true in spite of substances like piperine that aid passage through the gut wall. It is metabolized very quickly by the liver. Moreover, what is actually in a pill labeled as curcumin is highly variable, and curcumin is chemically unstable.

Many men rely on PSA to monitor prostate cancer progression. It may be misleading to use a supplement that may invalidate this important test. If there were any valid clinical studies indicating a true benefit, the corruption of a biomarker might be forgivable. But curcumin has only shown a benefit to mice so far. There are also some safety concerns (see this link). Patients must be wary of any supplement whose benefit is only supported by mouse/lab studies, and which only seems to affect PSA measurements. It is entirely possible to treat PSA without actually treating the cancer.

Wednesday, April 24, 2019

Should SBRT be THE preferred treatment for intermediate risk prostate cancer?

Last year, the American Society of Radiation Oncologists (ASTRO) looked at the available evidence comparing hypofractionated radiotherapy (either 60 Gy in 20 treatments or 70 Gy in 28 treatments) to standard fractionation (78-82 Gy in 40-44 treatments), and found it was at least as good in terms of oncological outcomes and toxicity. They found strong evidence for this recommendation (see this link). There are obvious benefits for the patient in terms of convenience and cost. They stopped short of strongly endorsing ultrahypofractionated radiation therapy (usually called SBRT), which is usually completed in only 4-5 treatments. There wasn't enough published data at the time.

Since then, there have been several published clinical trials, some with randomized comparisons. Jackson et al. have now compiled the data from 38 prospective clinical trials comprising 6,116 patients treated with SBRT for localized prostate cancer. Their meta-analysis found that 5-year biochemical recurrence-free survival (bRFS) was:

  • 97% among low-risk patients
  • 92% among intermediate-risk patients
  • more studies included intermediate risk than low risk
  • not enough high-risk patients to reliably report yet
  • 95% among all patients
  • 7-year bRFS was 94%
  • bRFS increased with higher doses
  • bRFS was not affected by the use of adjuvant ADT


In terms of physician-reported toxicity, they found:

  • Acute Grade ≥3 (serious) urinary toxicity occurred in 0.5% of patients
  • Acute Grade ≥3 (serious) rectal toxicity occurred in 0.1% of patients
  • Late-term Grade ≥3 (serious) urinary toxicity occurred in 2% of patients
  • Late-term Grade ≥3 (serious) rectal toxicity occurred in 1% of patients
  • Late urinary toxicity increased with dose, rectal toxicity did not


In terms of patient-reported adverse effects of treatment:

  • Urinary and Bowel scores returned to baseline within 2 years of treatment
  • They remained at those levels with 5 years of follow up
  • Sexual scores declined gradually over time


While the authors believe that their analysis provides enough evidence that SBRT should be considered a standard of care for low and intermediate risk patients, they stop short of recommending that SBRT be considered the standard of care for patients who choose radiotherapy.  (Active Surveillance is appropriate for most low risk patients.) There is an ongoing randomized clinical trial designed to prove whether SBRT or moderately hypofractionated radiation is superior. First results are expected in 2025. The PACE trials in the UK, will compare outcomes of SBRT vs surgery (PACE A) and SBRT vs IMRT (PACE B). Early toxicity results of PACE B have been presented. Results are expected in 2021.

Thanks to Amar Kishan for allowing me to see the full text of the analysis